You are on page 1of 28

Personality and Social Psychology Review Copyright © 2005 by

2005, Vol. 9, No. 3, 184-211 Lawrence Erlbaum Associates, Inc.

Embodiment in Attitudes, Social Perception, and Emotion

Paula M. Niedenthal
Laboratory in Social and Cognitive Psychology
CNRS and University of Clermont-Ferrand, France
Lawrence W. Barsalou
Department of Psychology
Emory University
Piotr Winkielman
Department of Psychology
University Qf California, San Diego
Silvia Krauth-Gruber and Fransois Ric
Laboratory in Social Psychology
Universite Rene, Descartes, Paris

Findings in the social psychology literatures on attitudes, social perception, and emo-
tion demonstrate that social information processing involves embodiment, where em-
bodiment refers both to actual bodily states and to simulations of experience in the
brain's modality-specific systems for perception, action, and introspection. We show
that embodiment underlies social information processing when the perceiver inter-
acts with actual social objects (online cognition) and when the perceiver represents
social objects in their absence (offline cognition). Although many empirical demon-
strations ofsocial embodiment exist, no particularly compelling account of them has
been offered. We propose that theories ofembodied cognition, such as the Perceptual
Symbol Systems (PSS) account (Barsalou, 1999), explain and integrate these find-
ings, and that they also suggest exciting new directions for research. We compare the
PSS account to a variety of related proposals and show how it addresses criticisms
that have previously posed problems for the general embodiment approach.

Consider the following findings. Wells and Petty various bodily positions associated nonobviously with
(1980) reported that nodding the head (as in agree- fear, anger, and sadness and found that these postural
ment) while listening to persuasive messages led to states modulated experienced affect. Strack, Martin,
more positive attitudes toward the message content and Stepper (1988) unobtrusively facilitated or inhib-
than shaking the head (as in disagreement). Caciop- ited the contraction of the zygomaticus (smiling) mus-
po, Priester, and Berntson (1993) observed that novel cle by asking participants to hold a pen in their mouth
Chinese ideographs presented during arm flexion (an while they evaluated cartoons. Participants judged car-
action associated with approach) were subsequently toons to be funnier when smiling was facilitated rather
evaluated more favorably than ideographs presented than inhibited (see Stepper & Strack, 1993, for related
during arm extension (an action associated with avoid- findings). Bargh, Chen, and Burrows (1996) showed
ance). Duclos et al. (1989) led participants to adopt that participants in whom the elderly stereotype had
been primed subsequently walked down a hallway
The authors thank Vic Ferreira, Art Glenberg, Danny McIntosh, more slowly than did participants in whom the stereo-
Randy O'Reilly, and Cathy Reed for their helpful comments on vari- type had not been primed. And Schubert (2004) showed
ous drafts of this article. We also thank the Society for Personality
and Social Psychology for awarding this article the SPSP 2003 The-
that making a fist influenced men's and women's auto-
oretical Innovation Prize. Preparation of this article was supported matic processing of words related to the concept of
by National Science Foundation grants BCS-0217294 to Piotr Winkiel- power.
man and BCS-0350687 to Piotr Winkielman and Paula Niedenthal. All such findings suggest that the body is closely
Requests for reprints should be sent to Paula M. Nieden- tied to the processing of social and emotional informa-
thal, LAPSCO, Universit6 Blaise Pascal, 34, avenue Carnot, 63037
Clermont-Ferrand Cedex, FRANCE. E-mail: niedenthal(srvpsy. tion. No single theory, however, has integrated the or to Piotr Winkielman, e-mail: pwinkiel@ findings or explained them in a unified manner. Recent theories of embodied cognition, which view knowl-

edge acquisition and knowledge use as processes sentations of the external world to the central unit. The
grounded in the brain's modality-specific systems, only function of the motor system is to dutifully exe-
hold promise of accounting for such findings and, per- cute the central executive's commands.
haps most important, predicting the effects explicitly Recent years have witnessed a crumbling of the first
and a priori (Barsalou, Niedenthal, Barbey, & Ruppert, claim of the computer metaphor-that of software-
2003; Smith & Semin, 2004). Further, these recent the- hardware independence. Research in cognitive and so-
ories are able to successfully address conceptual issues cial neuroscience has led to a growing appreciation that
that doomed previous embodiment proposals, making most phenomena are best understood by jointly con-
them attractive alternatives to widely accepted amodal sidering neural, psychological, and situational con-
theories of cognition. The aim of this article is to show straints (e.g., Brooks, 1991; Kosslyn, 1994; Winkiel-
how that is so and to propose new ideas for the study of man, Berntson, & Cacioppo, 2001). Nevertheless, the
information processing in social psychology. second claim of the computer metaphor lives on, and
many theories continue to assume that higher-order
cognition operates on amodal symbols. Noncontro-
The Notion of Embodied Mind versially, these theories assume that the actual experi-
ence of a current situation is initially represented in the
The nature of knowledge-the basic representa- brain's modality-specific systems. More controver-
tional elements of cognitive operations-lies at the sially, standard theories of cognition assume that the
core of psychology and cognitive science. Our view of modality-specific states experienced during an actual
what knowledge is determines how we conceptualize situation are redescribed and preserved in an abstract,
perception, memory, judgment, reasoning, and even amodal, language-like form, which we will refer to as
emotion. It is generally agreed that the processing of amodal symbols (Fodor, 1975). For example, on inter-
any mental content, including social and emotional acting with a particular individual, amodal symbols
content, involves internal symbols of some sort-men- redescribe the experienced perceptions, actions, and
tal representations. But this really just begs the ques- introspections to establish a conceptual representation
tion. What are mental representations? Further, how do of the interaction in long-term memory.
they derive their meaning?-an issue known as the As a person's knowledge about such interactions
symbol grounding problem (Harnad, 2003; Searle, grows, the underlying amodal symbols become orga-
1980). If we can make progress on these questions, we nized into structures that represent concepts extracted
can put psychology in general and social psychology in across experience (e.g., Collins & Quillian, 1969).
particular on firmer theoretical footing. These abstracted concepts constitute the person's
knowledge and allow the person to engage in infer-
ence, categorization, memory, and other forms of
Amodal Architectures higher cognition. Nearly all accounts of social cogni-
Most models guiding current cognitive and social tion represent knowledge this way, using feature lists,
psychology are based on the traditional computer met- semantic networks, schemata, propositions, produc-
aphor. This popular metaphor makes two major claims tions, frames, statistical vectors, and so forth, to re-
about the mind. The first is that the software of the describe people's perceptual, motor, and introspective
mind is independent of the hardware of the body and states (for discussions of such models, see Kunda,
the brain (Block, 1995; Dennett, 1969). Thus, cogni- 1999; Smith, 1998; Wyer & Srull, 1984). According to
tive operations are arbitrarily related to their physical all such views, amodal redescriptions of social experi-
instantiations so that any sufficiently complex physical ence constitute social knowledge.
system could have human intelligence. In principle, the The amodal architecture, although widely used, has
software that constitutes the mind (including the "so- recently been criticized on several grounds. One set of
cial mind") could run on anything-neurons, silicon, problems concerns the redescription process that pro-
or even wooden gears-as long as the elements were duces amodal symbols from modality-specific states in
arranged in proper functional relations. The second the first place. No direct empirical evidence exists for
claim of the computer metaphor is that high-level cog- such a process in the brain. Indeed, surprisingly few
nition, such as inference, categorization, and memory, theoretical accounts of this redescription process exist
is performed using abstract, amodal symbols that bear in the literature. More basically, there is no strong em-
arbitrary relations to the perceptual states that produce pirical case that the brain contains amodal symbols. In
them (Newell & Simon, 1972; Pylyshyn, 1984). Men- fact, arguments for amodal architectures are mostly
tal operations on these amodal representations are per- theoretical, based on assumptions about how cognition
formed by a central processing unit that is informa- should work, rather than on empirical evidence that it
tionally encapsulated from the input (sensory) and actually works this way. Further, as we discuss shortly,
output (motor) subsystems (Fodor, 1983). The only empirical findings increasingly challenge the basic as-
function of sensory systems is to deliver detailed repre- sumptions of the amodal architecture.

Given the lack of empirical evidence, why is the to take seriously the notion that knowledge is "em-
amodal architecture so widely accepted in both cog- bodied" or grounded in bodily states and in the
nitive and social psychology? There are a number of brain's modality-specific systems.' It is important to
important reasons. First, representations that employ note that the term "embodiment" has been used in
amodal symbols, such as semantic networks, feature multiple ways across the literature (Wilson, 2002).
lists, schemata, and propositions, provide powerful Many earlier embodiment theories emphasized the
ways of expressing the content of knowledge across role of actual bodily states in cognition. Examples of
various domains of knowledge, from perceptual im- such theories include Piaget's (1972) sensory-motor
ages to abstract concepts. Second, amodal symbols account of infant memory, and Zajonc and Markus's
provide a simple way to account for important func- (1984) hard-interface account of the interaction be-
tions of knowledge, such as categorization, categori- tween affect and cognition. In contrast, more contem-
cal inference, memory, comprehension, language, and porary embodiment theories emphasize simulations
thought (e.g., Anderson, 1983; Chomsky, 1959; New- of experience in modality-specific systems. Examples
ell, 1990; Newell & Simon, 1972). Third, amodal include Damasio's (1994) theory of emotion,
symbols have allowed computers to implement Glenberg's (1997) theory of memory, Barsalou's
knowledge. Because frames, semantic networks, and (1999) theory of perceptual symbol systems, and
property lists have many similarities to programming Gallese's (2003) theory of intersubjectivity. In the re-
languages, these representations can be implemented mainder of this article, we offer many specific exam-
easily on computers, not only for theoretical pur- ples and evidence for embodiment processes in both
poses, but also for applications (e.g., intelligent sys- peripheral (body-based) and central (modality-based)
tems in industry, education, and medicine). Fourth, senses of the term embodiment. However, as will be
until recently there were no compelling alternatives described in detail, our own theoretical perspective
that could account for the representation and function primarily focuses on the central sense of embodi-
of knowledge. For all these (good) reasons, amodal ment, or the brain's modality-specific systems. Those
approaches have dominated theories of representation systems include the sensory systems that underlie
for decades, even though little positive empirical evi- perception of a current situation, the motor systems
dence has accrued in their favor. Indeed, the theoretical that underlie action, and the introspective systems
virtues of amodal approaches have been so compel- that underlie conscious experiences of emotion, moti-
ling that it has not occurred to most researchers that vation, and cognitive operations.
seeking empirical support might be necessary. In- The main idea underlying all theories of embodied
stead, researchers typically assume that the amodal ar- cognition is that cognitive representations and opera-
chitecture is roughly correct and then go on from tions are fundamentally grounded in their physical
there to pursue their specific questions. context. Rather than relying solely on amodal ab-
Amodal models of knowledge are widespread in stractions that exist independently of their physical
social psychology. This is true despite the fact that instantiation, cognition relies heavily on the brain's
social psychology has provided some of the most modality-specific systems and on actual bodily states.
compelling evidence for what we present here as an One intuitive example is that empathy, or understand-
alternative view, namely, theories of embodied cogni- ing of another person's emotional state, comes from
tion. As we summarize later, experimental findings mentally "re-creating" this person's feelings in our-
consistent with embodiment theories abound in re- selves. The claim made by modern embodiment theo-
search on attitudes, empathy, and emotion. Thus, we ries is that all cognition, including high-level concep-
contend that continuing with the common assumption tual processes, relies heavily on such grounding in
of amodal representation will lead us down a false either the modalities or the body (Wilson, 2002). This
path and that advances in the understanding of claim is significant given that embodiment theories
knowledge, in general, and social knowledge, in par- have traditionally been viewed as having little to say
ticular, can be made if social psychology starts to about higher cognitive functions, not just empathy,
question the amodal architecture or at least looks but also abstract concepts, categorical inference, and
elsewhere for further inspiration. the ability to combine internal symbols in novel, pro-
ductive ways. As we will see, theories of embodied
Embodied Architectures cognition are increasingly able to explain how such
phenomena can be based in modality-specific sys-
In recent years, researchers in psychology (Bar- tems and bodily states.
salou, 1999; Glenberg & Robertson, 2000; Parsons et
al., 1995), philosophy (Churchland, Ramachandran,
Some of the more recent philosophical predecessors of embodi-
& Sejnowski, 1994; Clark, 1997; Prinz, 2002; Varela, ment theories can be found in writings of Ryle (1949), Merleau-
Thompson, & Rosch, 1991), robotics (Brooks, 1991), Ponty (1963), and Heidegger (1962). For further discussion, see
and linguistics (Lakoff & Johnson, 1999) have started Prinz (2002).


Embodiment Effects of the previously met tall and imposing individual, can
in Cognitive Psychology produce embodied responses in the perceiver that un-
derlie representation of the symbol's meaning. For ex-
Recent studies in cognitive psychology have detn- ample, upward head orientation and defensive bodily
onstrated that conceptual knowledge is embodied (we responding might implicitly contribute to the infer-
address the social literature shortly). As we review ences that the individual is tall and imposing. A strong
some of these studies, we ask the reader to note two embodiment view argues that the modality-specific
things. First, note that these effects cannot be easily states engaged in during online cognition constitute the
predicted a priori by amodal theories, although as Bar- knowledge that is acquired and later used in offline
salou (1999) notes those theories can explain any effect cognition. According to this view, stored embodiments
post hoc by adding increasingly complex assumptions constitute the basic elements of knowledge. To es-
about representation and processing. Second, note that tablish the meaning of symbols during offline process-
whereas some studies we cite only show a correlation ing, people rely on repertoires of modality-specific re-
between conceptual operations and modality-specific sponses acquired previously during online processing
systems, others provide direct causal evidence. The of these symbols' referents.
correlational results are useful because they confirm a
priori predictions derived from the embodiment ac-
count. But, it is increasingly essential to demonstrate Online Effects
the causal roles of embodiment in higher cognition. The idea that modality-specific processes partici-
Fortunately, as we will see, many studies experimen- pate in the conceptual processing of real world objects
tally manipulate embodiments across randomly as- can be illustrated with research on the compatibility
signed groups of participants, thereby demonstrating between motor actions and conceptual tasks. Tucker
causal effects. and Ellis (1998) asked participants to detect whether a
cup was right side up or upside down. Although the
handle of the cup was irrelevant to the judgment, par-
Online Embodiment ticipants responded faster when the cup's handle was
and Offline Embodiment on the same side of the display as the response hand
Wilson (2002) distinguished between online and than when the handle was on the opposite side. This re-
offline embodiment. The term online embodiment, and sult indicates that representations of possible actions
the related term, situated cognition, refer to the idea (e.g., reaching for the cup) influence a perceptual judg-
that much cognitive activity operates directly on real- ment even when these actions are not relevant to the
world environments. Accordingly, cognitive activity is judgment. Reed and Farah (1995) asked participants to
intimately tied to the relevant modality-specific pro- judge whether two human figures depicted the same
cesses required to interact with the environment effec- posture. Participants asked to move their own arms
tively. For example, when meeting a new individual performed relatively better at detecting changes in the
(e.g., a tall and imposing person), a perceiver spontane- arm position of a visually presented figure, whereas
ously produces in vivo sensory and somatic responses participants asked to move their own legs did relatively
(e.g., looking up and feeling apprehensive) as well as better at detecting changes in the figure's legs. Again,
motor responses (e.g., stepping back to keep distance). this finding suggests that representations of partici-
The embodiment account views these sensory, somat- pants' own bodies contribute to the performance on the
ic, and motor responses as necessary for the encoding visual task.
and interpretation of the new individual, not simply The just described behavioral studies are consistent
as a by-product of a purely amodal analysis. Another with neuroimaging research that found activation of the
useful way to conceptualize online embodiment is as grasping circuit when participants viewed manipulable
knowledge acquisition, with the perceiver acquiring objects while lying passively in an fMRI scanner (Chao
and modifying a repertoire of modality-specific re- & Martin, 2000). Related research with monkeys shows
sponses to stimuli as he or she interacts actively with that motor neurons involved in controlling tool use fire
the social environment (also see Gallese, 2003). A cen- when the tools are merely perceived and no motor re-
tral tenet of recent theories is that the establishment of sponse is possible (Rizzolatti & Arbib, 1998).
this repertoire plays a central role in higher cognition.
The term offline embodiment refers to the idea that Offline Effects
when cognitive activity is decoupled from the real-
world environment, cognitive operations continue to Numerous studies have also documented offline
be supported by processing in modality-specific sys- embodiment or the involvement of modality-specific
tems and bodily states. Just thinking about an object states when processing is decoupled from the environ-
produces embodied states as if the object were actually ment (as when the object is absent or represented
there. Thus, perceiving a symbol, for example the name solely by a symbol, such as a word or a picture). For

example, in a study by Rauscher, Krauss, and Chen Churchland et al. (1994), Clark (1997), Glenberg
(1996), participants first watched an animated action (1997), Lakoff and Johnson (1999), Prinz (2002), and
cartoon. After a break, with the cartoon no longer pres- Varela, Thompson, and Rosch (1991).
ent, participants were then asked to describe the car-
toon to a listener. When participants were prevented
from gesturing (under the guise of recording skin con- Embodiment Effects
ductance from their palms), they were significantly in Social Psychology
slower to describe spatial elements of the cartoon. Pre-
sumably, blocking the embodiment impaired access to Our primary argument here is that social psychol-
the conceptual elements of the representation. In an- ogy could profit from theories of embodied cognition.
other example, Spivey and his colleagues report that In particular, such theories can help integrate and ac-
participants who listen to vignettes including spatial count for the findings listed willy-nilly at the beginning
descriptions, such as "the top of a skyscraper" or "the of this article, as well as those just described. Further-
bottom of a canyon," perform appropriate eye move- more, these theories can help us generate interesting
ments up or down, respectively, as if actually present in predictions that cannot be derived a priori from the
the situation (Spivey, Tyler, Richardson, & Young, amodal accounts of knowledge representations that
2000). Finally, Glenberg and Kaschak (2002) found currently dominate social psychology. To support our
that participants were faster at judging the sensibility argument, we next summarize findings that illustrate
of a sentence when its meaning was compatible with embodiment in three traditional areas of social psy-
the hand movement required for the response (e.g., chology: attitudes, social perception, and emotion.
"Close the drawer"-forward movement; "Open the Then, we present a specific theory of embodiment,
drawer"-backward movement). Remarkably, this ac- Barsalou's (1999) PSS account of conceptual process-
tion-sentence compatibility effect occurred even when ing. We compare PSS to some views already present
the sentences referred to abstract actions that involved in the social psychology and related literatures and ex-
directional communication (i.e., participants were fast- plain how this account deals with prior criticisms of
est in judging the sensibility of the sentence "You the embodiment approach. Finally, we show how
told Liz the story" with a forward movement and the PSS sheds new light on classic phenomena in social
sentence "Liz told you the story" with a backward psychology.
movement). Richardson, Spivey, Barsalou, and McRae In the following sections, we organize embodiment
(2003) report an analogous set of findings. findings around the themes of attitude, social percep-
Again, these behavioral studies are consistent with tion, and emotion. Within each group of findings, we
neuroscientific data. In the brain lesion literature, distinguish again between online embodiment that oc-
many studies have found high-level cognitive impair- curs in the presence of real external stimuli and offline
ments as a result of neurological damage to modal- embodiment that occurs during the use of symbols that
ity-specific systems. Lesions in these systems produce refer to real stimuli not actually present. For example,
systematic deficits in category knowledge (e.g., Cree imitation of another person's happy facial expression is
& McRae, 2003; Damasio & Damasio, 1994; Farah, an example of online embodiment. On the other hand,
1994; Humphries & Forde, 2001; Simmons & Bar- understanding the word "happiness" or recalling a hap-
salou, 2003; Warrington & Shallice, 1984). Lesions in py experience by recruiting modality-specific systems
modality-specific areas also produce deficits in the is an example of offline embodiment. We find the on-
representation of episodic memories (e.g., Rubin & line/offline distinction useful in organizing the so-
Greenberg, 1998). Recently, activation of modality- cial psychology literature, in part because it can serve
specific areas has been observed when people perform as a way to conceptualize knowledge acquisition and
abstract conceptual tasks, such as concept property knowledge use and to see similarities in their underly-
verification, that require deep, nonassociative process- ing mechanisms. Before we start, we hasten to add that
ing of target stimuli (Kan, Barsalou, Solomon, Minor, our summaries of these empirical literatures should by
& Thompson-Schill, in press). no means be viewed as exhaustive. We simply try to
In sum, accumulating evidence from cognitive psy- highlight findings that are representative for each cate-
chology and cognitive neuroscience supports embodi- gory (see Barsalou et al. 2003, for additional discus-
ment theories of knowledge. For more extensive re- sion of these literatures).
views of such findings, see Martin (2001), Barsalou
(2003b), and Hegerty (2004). Importantly, such find-
ings are not predicted a priori by amodal accounts. Ac- Embodiment of Attitudes
cordingly, these results are increasingly shaping theo- Charles Darwin (1904) defined attitude as a col-
rizing in cognitive psychology and cognitive science, lection of motor behaviors-especially posture-that
as well as in philosophy and linguistics. For examples convey an organism's affective response toward an ob-
of these theoretical accounts, see Barsalou (1999, 2003a), ject. Thus, it would not have come as any surprise to

him that the body is involved in the acquisition and use action typically associated with approach, reflective
of attitudes. Subsequent accounts similarly stressed the of positive attitudes) or to push downward on the table-
importance of motor behavior in attitudes (e.g., Sher- top (an action typically associated with avoidance, re-
rington, 1906; Washburn, 1926). Francis Galton flective of negative attitudes) while they were exposed
(1884), for example, also defined attitude in terms of to the ideographs. Consistent with an embodiment
posture (literally, as a bodily inclination). He believed hypothesis, ideographs seen during the approach be-
that the way to quantify a person's attitude toward an- havior were later rated more positively than were ideo-
other individual was to sit the two individuals in adja- graphs seen during the avoidance behavior. A sub-
cent chairs and then measure the weight that they ap- sequent study in which a control group (that engaged in
plied to the edge of the chair nearest to the other no behavior) was added to the experimental design
individual (vs. the back of the chair). Individuals who demonstrated that both approach and avoidance behav-
like one another should put more weight on the edge of iors had a significant influence on attitudes.
the chair facing the other person, he argued, thus mani- As all these studies illustrate, bodily postures and
festing their positive attitude toward each other. Iron- motor behavior are associated with positive and nega-
ically, because a focus on posture no longer figures into tive inclinations and action tendencies toward objects.
the definition of attitude, the following studies may Furthermore, these inclinations and tendencies influ-
have more novelty value now than 100 years ago. ence attitudes toward those objects as expressed by
self-report and attitude ratings. Thus, attitudes appear
Online embodiment in the acquisition and pro- to be determined, at least in part, by embodied re-
cessing of attitudes. The studies we summarize in sponses. We next look at the role of embodiment when
this section suggest that bodily responses during inter- attitude objects are not present.
action with novel objects influence later-reported atti-
tudes and impressions. In an early demonstration of Offline embodiment in attitude processing. As
such an effect, Wells and Petty (1980) instructed par- mentioned previously, the notion of offline embodi-
ticipants to nod their heads vertically or to shake their ment is that modality-specific systems are engaged
heads horizontally while wearing headphones, under even when people process symbolic entities, such as
the pretext that the research was designed to investigate words. This is because conceptual processing draws on
whether the headphones slipped off as listeners moved the modality-specific patterns established earlier dur-
to the music. While nodding or shaking their heads, ing online acquisition processing. Furthermore, the
participants then heard either a disagreeable or an embodiment view proposes that conceptual processing
agreeable message about a university-related topic. is maximally efficient when relevant conceptual infor-
Later, they rated how much they agreed with the mes- mation is consistent with current embodiments.
sage. Wells and Petty found that the earlier head move- This hypothesis is supported by the findings of
ments later modulated participants' judgments. Spe- Chen and Bargh's (1999) study in which participants
cifically, participants who had nodded while hearing were exposed to words with positive or negative va-
the message were more favorable than participants lence (e.g., love, hate) and had to report the valence by
who had shaken their heads. pulling a lever toward them or by pushing it away. Con-
Tom, Pettersen, Law, Burton, and Cook (1991) ex- sistent with an embodiment prediction, participants
tended this study and induced participants to nod their made pulling responses faster when responding to pos-
heads (in agreement) or to shake their heads (in dis- itive words, compared to negative, and made pushing
agreement) while placing a pen on the table in front of responses faster to negative words compared to posi-
participants. After the purported testing of the head- tive ones. In a second study, Chen and Bargh had par-
phones, a naive experimenter offered to give the partic- ticipants indicate when a word merely appeared on
ipant the "old" pen that had been placed on the table the computer screen-participants made the same re-
during the experiment or a "new" pen that the partici- sponse to all words regardless of their affective va-
pant had never seen. Individuals who had nodded dur- lence. Subjects who indicated a word's appearance by
ing the testing of the headphones preferred the old pen, pulling the lever toward them responded faster to posi-
whereas participants who had shaken their heads pre- tive words than to negative ones. Participants who indi-
ferred the new one. Presumably, whether participants cated a word's appearance by pushing the lever away
nodded or shook their heads during initial exposure to responded faster to negative words. Thus, there was a
the pen influenced the attitude that they developed to- systematic relationship between the processing of the
ward it, as revealed in their later preference. word and the compatibility between the valence of the
Cacioppo et al. (1993) explored the relation be- word and the behavior used in response to it (see also
tween a different attitude-relevant motor behavior and Neumann & Strack, 2000; Wentura, Rothermund, &
the development of attitudes toward completely novel Bak, 2000).
stimuli-Chinese ideographs. Participants were in- Forster and Strack (1997, 1998) demonstrated a
duced to push upward on a table from underneath (an similar effect in the retrieval of information from long-

term memory. Participants in their study generated the unconscious enactments of thoughts and memories
names of famous people and later classified the people involving other significant people (Breuer & Freud,
according to whether they liked, disliked, or were neu- 1983-95/1955). And in a more general embodiment ac-
tral about them. During the name generation task, par- count, psychoanalyst Felix Deutsch (1952) proposed
ticipants either pulled up on the table in front of them that "all automatic ... movements represent in some
from underneath its bottom surface (an approach be- way the search for a desired ... [person] ... from the
havior, as described earlier) or pushed down on its past" (p. 210). He argued, based on clinical cases, that
top surface (an avoidance behavior). Participants who parts of the body actually personify members of one's
performed the approach behavior during name genera- family (even, or especially, when they are not physi-
tion retrieved more names of people they liked. Con- cally present) and that sensations and movements in
versely, participants who performed the avoidance ac- those body parts express feelings toward and memories
tion retrieved more names of people they disliked. of those people. These unique perspectives have rarely
Thus, participants' motor behavior influenced the re- been integrated explicitly with empirical research on
trieval of attitude objects from long-term memory in an social perception and impression formation. Still, the
attitude-congruent manner. social psychology literature is replete with examples
In sum, the studies described in this section on atti- that are consistent with such a view.
tudes demonstrate the two embodiment effects of inter-
est. First, during online exposure to objects, the pro-
duction of motor movements associated with positive Online embodiment in social information pro-
attitudes leads to the later expression of positive at- cessing: Mimicry and imitation. Research has con-
titudes, and the production of motor movements as- sistently shown that perceivers imitate the facial ges-
sociated with negative attitudes leads to the later ex- tures of perceived others. O'Toole and Dubin (1968)
pression of negative attitudes. Second, during offline demonstrated that mothers open their mouths in re-
cognition, processing symbols that stand for absent at- sponse to the open mouth of their infant who is about to
titude objects are most efficient when a congruent mo- feed. These imitative behaviors occur very early in de-
tor behavior is maintained, suggesting that represent- velopment. In classic studies, Meltzoff and Moore
ing the conceptual knowledge involves the relevant (1977, 1989) showed that neonates imitate basic facial
motor behavior.2 gestures such as tongue protrusion and mouth opening,
suggesting a biological basis of basic imitation skills
Social Perception3 (for a review, see Meltzoff & Prinz, 2002). The biologi-
cal argument is strengthened further by observations of
It may feel intuitively correct to learn that individu- basic imitative behaviors in other primates (Preston &
als embody the behaviors of others online or when de Waal, 2002) and by the impairments of basic imi-
those others are physically present. Researchers have tation as a result of developmental disorders such as
long argued for the role of mimicry and imitation in so- autism (McIntosh, Reichmann-Decker, Winkielman,
cial modeling, social coordination, and empathy (e.g., & Wilbarger, 2004; Rogers, 1999; Sigman, Kasari,
Bandura, 1977; Lipps, 1907). It is perhaps more coun- Kwon, & Yirmiya, 1992).
terintuitive to imagine how embodiment enters into so- Importantly, imitation extends beyond facial behav-
cial perception offline, or when other people are pres- ior. Individuals engaged in conversation tend to syn-
ent only symbolically. This notion has some interesting chronize their latency and rate of speech, the duration
historical precursors. Freud, of course, thought that of their utterances, and other speech characteristics
the sensory-motor symptoms of hysterics were in fact (e.g., Capella & Planalp, 1981; Matarazzo & Wiens,
1972; Webb, 1972). Listeners also tend to mimic talk-
2As we discuss later, the embodiment view allows for substantial ers' emotional prosody (e.g., Neumann & Strack,
flexibility in terms of what specific modality states and physical 2000), manual gestures (e.g., Bavelas, Black, Chovil,
movements are associated with specific concepts. For example, un- Lemery, & Mullett, 1988; Maxwell, Cook, & Burr,
der different task settings, positive versus negative concepts might be 1985), and even their syntactic constructions (e.g.,
associated with different bodily movements (push or pull, up or Bock, 1986). Much research has focused on postural
down, toward or away). The main message of the embodiment view
is that conceptual representations are supported by simulations in synchrony. For example, in one study Bernieri (1988)
modality systems, not that they are rigidly tied to specific bodily had judges code the postures of two individuals filmed
states. This view is consistent with recent findings on flexibility of while they were actually interacting with each other
the link between valence and specific bodily movement (Markman & and the same two individuals who appeared to be inter-
Brendl, in press). acting with each other but who were actually interact-
3 Niedenthal and Halberstadt (2004) have argued that the term
social perception as used in social psychology is a misnomer, be-
ing with different people. Supporting the idea of imita-
cause rarely have perceptual processes actually been examined. tion, the results revealed greater postural synchrony for
Here, similar to most other researchers, we use this term broadly in two individuals engaged in actual interaction than for
referring to impression formation and social information processing. two individuals in a contrived interaction (for related

results, see Bernieri, Reznick, & Rosenthal, 1988; Ber- tions that share, with language, the propositional for-
nieri & Rosenthal, 1991). In a more recent study by mat. This enables one to ascribe others' intentions,
Chartrand and Bargh (1999), a trained experimenter desires and beliefs, and therefore to understand the
mental antecedents of their overt behavior ... [but this
rubbed her nose or shook her foot while she interacted is a] disembodied view ... I think that there is now
with participants. When the experimenter scratched enough empirical evidence to reject a disembodied
her nose, participants were more likely to scratch their theory of the mind as biologically implausible.
nose (than to shake their foot). And when the experi-
menter shook her foot, participants were more likely to
shake their foot (than to scratch their nose). This result Offline embodiment in social information pro-
further suggests that the mere observation of another cessing: Category priming and motor responding.
person engaging in a particular action facilitates the In a recent and already classic series of studies, Bargh,
same action in the perceiver. Chen, and Burrows (1996) demonstrated embodiment
It is widely believed that synchrony facilitates coop- in social information processing when the actual social
eration and empathy among interaction partners (e.g., stimuli were not present. Participants in one study were
Hatfield, Cacioppo, & Rapson, 1993; LaFrance, 1985; instructed to form sentences from groups of words pre-
LaFrance & Ickes, 1981; Neumann & Strack, 2000; sented in random order. In the critical conditions, many
Semin, 2000). Consistent with this belief, enhancing of the sentences contained words related to the stereo-
mimicry increases smoothness of interaction and lik- type of the elderly (e.g., gray, Florida, and bingo).
ing between partners (Chartrand & Bargh, 1999). It is Importantly, those words were not specifically related
also believed that interpersonal closeness facilitates to motor movement. In the control conditions, the sen-
mimicry (Bush, McHugo, & Lanzetta, 1989; McIn- tences that participants constructed contained neutral
tosh, Druckman, & Zajonc, 1994; van Baaren, Mad- words unrelated to the elderly stereotype. Results
dux, Chartrand, de Bouter, & Knippenberg, 2003). showed that when participants had been primed with
Consistent with this assumption, participants show the elderly stereotype, they actually took longer to
more spontaneous mimicry of a model's behavior walk from the experimental room to the elevator than
when their liking for the model is experimentally en- did control participants. Presumably, this occurred be-
hanced or when the model is the participant's friend cause the priming task activated the elderly stereotype
(McIntosh, in press). which contains knowledge that old people tend to
A specific neural mechanism appears to underlie move slowly. In turn, this knowledge activated action
imitation. Rizzolatti and his colleagues have observed schemas, which caused the embodiment effect of
that certain circuits involved in the production of an or- slow walking. Studies in the same research project
ganism's motor behavior also become active in re- have demonstrated embodiment effects of processing
sponse to perceived intentional motor behavior (e.g., other social categories on other kinds of behavior, in-
Rizzolatti, Fadiga, Fogassi, & Gallese, 2002; also see cluding rudeness and aggressiveness (for a review, see
Chao & Martin, 2000). Such mirror neuron circuits Dijksterhuis & Bargh, 2001).
could play two important adaptive roles in social life. Another class of embodiment effects in offline so-
First, they may support fast learning, such that an or- cial perception occurs during activation of evaluative
ganism learns new actions through imitation (Gallese, knowledge. In one demonstration, Vanman, Paul, Ito,
2003). This is consistent with our idea of online em- and Miller (1997) instructed participants to form im-
bodiment as knowledge acquisition. Second, these cir- ages of people with whom they might later work on a
cuits may be responsible for social contagion, such as problem-solving task. A number of different variables
the induction of congruent emotional states in others, moderated participants' facial responses, as measured
especially if those others are psychologically close by EMG. For example, participants were most likely to
(Decety & Chaminade, 2003). If so, then these circuits display positive facial reactions when their imagined
should be centrally implicated in empathy and social partners, who were competent (vs. incompetent), ex-
cooperation (Hatfield, Cacioppo, & Rapson, 1992). erted high (vs. low) effort, or belonged to the same (vs.
More generally, the growing evidence for the close different) racial category. Thus embodiment occurred
involvement of mirror neurons in empathy, imitation, when individuals activated representations of people
and attribution of mental state supports the embodi- who were not actually present.
ment view. Reflecting on this evidence, Gallese (2003) Finally, Andersen, Reznik, and Manzella (1996) ob-
noted that according to amodal accounts of social tained personality descriptions about significant others
cognition, in their participants' lives, and then developed descrip-
tions of fictional characters who partially resembled
when faced with the problem of understanding the them. In a later experimental session, participants read
meaning of others' behaviors, adults humans must the descriptions of these fictional characters, not real-
necessarily translate the sensory information about the izing that they were in any way related to their signifi-
observed behavior into a series of mental representa- cant others. In line with an embodiment account, par-

ticipants' facial expressions, as coded by a naive judge, tion stimuli. Evidence is accumulating that people
were influenced by what they read. When participants mimic others' emotional facial expressions (Bush,
read about characters based on significant others they Barr, McHugo, & Lanzetta, 1989; Dimberg, 1982). In
liked, they tended to produce positive facial expres- Bavelas, Black, Lemery, and Mullett (1986), for exam-
sions. Conversely, when participants read about char- ple, a confederate faked an injury and then grimaced
acters based on significant others who they disliked, in pain. When participants observed the grimace, they
they tended to produce negative facial expressions. grimaced themselves. Furthermore, the magnitude of
Thus, again, simply reading about abstract social stim- participants' grimaces increased with how clearly they
uli influenced facial responding, suggesting an embod- could see the confederate's grimace. Emotion imi-
ied representation of social knowledge. tation appears to be relatively automatic and to even
To conclude this section, we have described illustra- be elicited outside awareness, as when participants
tive studies showing, first, that individuals tend to react with slight smiles and frowns to subliminal
mimic the motor behavior of other individuals when happy and angry expressions (Dimberg, Thunberg, &
those others are actually present and, second, that em- Elmehed, 2000). Further evidence suggests that em-
bodied responses are engaged when individuals ma- bodied consequences of subliminal facial expressions
nipulate information offline about other people stored extend beyond facial mimicry. In one study, for exam-
in long-term memory. Recent findings in social neur- ple, participants were first subliminally exposed to
oscience also provide strong evidence for this inter- happy or angry faces and were then asked to try a novel
pretation. All these findings are closely related to the beverage. The results showed that participants exposed
research findings on emotion and empathy that we ad- to subliminal happy faces later behaved more in an
dress next. approach-oriented fashion (by pouring and drinking
more beverage) than subjects who were exposed to
Emotion subliminal angry faces (Winkielman, Berridge, & Wil-
barger, in press).
Students of emotion most often associate the idea of According to embodiment views, bodily responses
embodiment with William James (1890). According to should facilitate cognitive processing of emotion stim-
James, the basis of emotion is the bodily activity that uli. In one demonstration of this effect, Wallbott (199 1)
occurs in response to an emotional stimulus. Thus, had participants categorize the emotional facial ex-
James was claiming that emotions are embodiments.4 pressions displayed in photographs of other people. As
Our goal here is not to assess whether everything about participants categorized the photographed expressions,
James's theory is correct. In particular, we are not en- their own faces were surreptitiously videotaped. Re-
dorsing James's claim about the necessity of the auto- sults showed that the participants tended to mimic
nomic nervous system for emotion. We are also not en- the facial expressions as they categorized them. When
dorsing a strong mapping between specific emotions they categorized happy faces, for example, they smiled
and specific embodiments. We do suggest, however, themselves. Furthermore, participants' accuracy in
that embodiment is critically involved in information classifying the facial expressions was positively corre-
processing about emotion-not only "online," when lated with the extent of mimicry. The more participants
people respond to real emotion objects, but also "off- mimicked the faces, the better they were at discerning
line," when people represent the meanings of emo- what expression the face was displaying.
tional symbols, such as words. Furthermore, as de- Niedenthal, Brauer, Halberstadt, and Innes-Ker
scribed shortly, we propose a concrete embodiment (2001) demonstrated that this mimicry plays a causal
account grounded in current research in psycholo- role in the processing of emotional expression. Partici-
gy and neuroscience that allows us to make specific pants watched one facial expression morph into an-
predictions about the role of embodiment in emotion other and had to detect when the expression changed.
phenomena. Some participants were free to mimic, whereas others
were prevented from mimicking by holding a pencil
Online embodiment in emotion processing. We laterally between their lips and teeth. Consistent with
have already discussed the ubiquity of embodied re- the embodiment hypothesis, participants free to mimic
sponses to nonemotional actions and gestures. In the expressions detected the change in emotional ex-
this section, we focus on embodied responses to emo- pression earlier (more efficiently) for any facial ex-
pression than did participants who were prevented
from mimicking the expressions (for further discussion
4More specifically, James argued that the conscious experience see Niedenthal, Ric, & Krauth-Gruber, 2002).
of emotion (the subjective feeling component of emotion) derives
from the conscious perception of embodiments. Note, however, that
Adolphs, Damasio, Tranel, Cooper, and Damasio
emotion can be embodied without these embodiments being con- (2000) report further evidence for the causal involve-
sciously represented as feelings (for discussion, see Berridge & ment of somatosensory processes in recognition of
Winkielman, 2003). facial expressions. Clinical patients with lesions in

somatosensory cortex showed poorer performance in 2003). All these findings were interpreted as evidence
classifying facial expressions than individuals without of an embodied simulation in the perceiver of what was
such lesions. Presumably, simulating emotional ex- happening to the perceived person (for summaries of
pressions on one's own face, and experiencing the re- related research, see Gallese, 2003; lacoboni, in press).
sulting somatosensory feedback, is necessary for the In short, there is now converging evidence that em-
process of recognition.5 In short, the results reported bodiment, in the sense of actual motor and soma-
by Wallbott (1991), Niedenthal et al. (2001), and tosensory responses, is involved in empathy.
Adolphs and his colleagues all converge on the conclu-
sion that feedback from facial mimicry is importantly Offline embodiment in emotion. As discussed
involved in a perceiver's ability to process emotional earlier, offline embodiment is the use of modality-spe-
expressions. Carr, Jacoboni, Dubeau, Mazziotta, and cific representations to represent the meaning of sym-
Lenzi (2003) have begun to explore the neural circuit bols whose referents are absent. Offline embodiment
that underlies this process. often appears central to the representation of emotion
We have noted several times that the mimicry of emo- knowledge. In a study by Laird, Wagener, Halal, and
tional gestures has been proposed as a mechanism that Szegda (1982), participants studied both anger-pro-
supports empathy (Lanzetta & Englis, 1989; McIntosh voking and happiness-provoking material. Later, par-
et al., 1994; Vaughan & Lanzetta, 1980; see Levenson, ticipants were covertly induced to smile or frjown and
1996 for discussion). Zajonc, Adelmann, Murphy, and then were instructed to recall the earlier learned mate-
Niedenthal (1987) further demonstrate this relation- rial. Results showed that the induced expression mod-
ship. These researchers compared the facial similarity erated recall. Participants induced to smile recalled the
of couples at the time of their marriage to their facial happy material better than those induced to frown,
similarity after 25 or more years of marriage. Zajonc et whereas participants induced to frown recalled the an-
al. reasoned that married partners should frequently gry material better than those induced to smile. Impor-
mimic each other's facial expressions, presumably be- tantly, this effect was found only when participants' fa-
cause they are particularly motivated to empathize with cial expressions were accompanied by a congruent
each other. As a consequence of this frequent mimicry, emotional state. That is, participants' memory perfor-
the couples' faces should grow more similar over time. mance was maximized when the motor behavior, the
Consistent with this reasoning, Zajonc et al. found that emotional state, and the emotional meaning of the
after 25 or more years of marriage, facial similarity be- learned material were all compatible (see related stud-
tween couples was greater than at the time of their mar- ies by Strack et al., 1988; Zajonc, Pietromonaco, &
riage and also greater than between random people of Bargh, 1982).
the same age. Furthermore, this effect was correlated In a similar study by Riskind (1984), participants
with the quality of the marriage and therefore presum- were instructed to retrieve pleasant or unpleasant auto-
ably success in empathizing. biographical memories while they adopted different
The rather indirect finding of Zajonc and his col- postures and facial expressions. The embodiment manip-
leagues (1987) is supported by recent studies on the ulation was expected to influence the emotional nature
neural basis of mirroring effects. One fMRI study ob- of the memories recalled. As predicted, postural and fa-
served very similar changes in brain activity of a fe- cial manipulations modulated the latencies to retrieve
male participant while painful stimulation was applied positive versus negative life experiences. Adopting an
to her own hand or to her partner's hand (Singer et al., erect posture and smiling speeded the retrieval of pleas-
2004). A related study used single cell recording and ant autobiographical memories, relative to the speed of
found activation of pain-related neurons when a pain- retrieving unpleasant autobiographical memories.
ful stimulus was applied to the participant's own hand Research by Stepper and Strack (1993) generalized
and when the patient watched the painful stimulus ap- these effects to how people respond emotionally to
plied to the experimenter's hand (Hutchison, Davis, evaluations of their performance. Participants were led
Lozano, Tasker, & Dostrovsky, 1999). Yet another to sit in an upright or slumped position under the pre-
study found activation in the insula (a brain area re- text that the experimenters were concerned with task
sponsible for processing somatosensory information) performance under varying ergonomic conditions.
when the participant was exposed to disgusting odors While upright or slumped, participants performed an
and when the participant simply watched a movie of achievement test and received bogus feedback that
other people's expressions of disgust (Wicker et al., they had done well. Later, participants rated their feel-
ing of pride at the time. Participants who had sat up-
right while receiving task feedback reported experi-
5Somatosensory mechanisms are not only involved in recogniz- encing more pride than participants who had sat in a
ing facial expressions. Similar effects of damage to the somatosen-
sory cortex have also been obtained with tasks requiring emotion slumped position (see also Riskind & Gotay, 1982).
recognition from prosody and body movement (Adolphs, Damasio, To summarize, these findings provide strong evi-
& Tranel, 2002; Heberlein, Adolphs, Tranel, & Damasio, in press). dence for the embodiment of emotion processing. This

evidence also suggests that embodiment supports im- address how such representations are used in the simu-
portant cognitive and social functions, such as recogni- lations that underlie conceptual processing.
tion, memory, empathy, and understanding. As such,
the research on emotion embodiment is consistent with Feature Maps and Convergence Zones
earlier reviewed research on embodiment in social per-
ception and attitudes and points to the critical impor- The PSS account takes as a starting point Damasio's
tance of modality-specific states in the representation (1989) theory of convergence zones (CZ) proposed by
and processing of social knowledge. But just how does Damasio and his colleagues (see Simmons & Barsalou,
this work? 2003, for an elaborated account). CZ theory assumes
that the perception of an object activates relevant fea-
ture detectors in the brain's modality-specific systems.
An Integrative Theory of Embodiment The populations of neurons that code featural informa-
tion in a particular modality are organized in hierarchi-
For over 30 years, evidence has accumulated that cal and distributed systems of feature maps (Palmer,
implicates embodiment centrally in the acquisition and
expression of attitudes, in social perception, and in the
1999; Zeki, 1993). When a stimulus is perceived on a
learning and use of emotion knowledge. Despite all given modality, populations of neurons in relevant
this evidence, however, no major theory has explained maps code the stimulus' features on that modality in a
hierarchical manner. For example, visual processing of
it (Smith & Semin, 2004). Furthermore, the common a happy face activates feature detectors that respond to
interpretation of such findings is that embodiment ef- the color, orientation, and planar surfaces of the face.
fects arise "simply" as peripheral by-products of con-
Whereas feature detectors early in the processing
ceptual knowledge, which is typically viewed as the stream code detailed perspective-based properties of
critical cause of social cognition. Important accounts the face, higher-order detectors code its more abstract
of the relationship between perception and action have and invariant properties. The pattern of activation
been proposed in recent years (e.g., Bargh & Char-
across relevant features maps represents the face in vi-
trand, 1999; Neumann, Fbrster, & Strack, 2003). These sual processing.
accounts, however, have emphasized the direct and au-
tomatic nature of the relation between perceptual and Analogously, CZ theory assumes that systems of
feature maps reside in the other sensory-motor modal-
motor processes. In contrast, we explicitly focus on the
ities and in the limbic system for emotion. All these
dynamical role of modality-specific systems in repre-
senting and manipulating conceptual knowledge. In maps operate in parallel, so that while a face is being
what follows, we describe the PSS theory, which puts represented in visual feature maps, sounds produced
embodiment at the core of information processing, by the face are being coded in auditory feature maps,
affective responses to the face are being coded in
including attitudes, social perception, and emotion. limbic feature maps, bodily responses to it are being
Comprehensive presentation of PSS can be found else- coded in motor feature maps, and so forth.
where (e.g., Barsalou, 1999, 2003a, 2003b). Our pur-
CZ theory further proposes that conjunctive neu-
pose here is to describe the account in enough detail to rons in the brain's association areas capture and store
show how it explains embodiment effects in social psy- the patterns of activation in feature maps for later rep-
chology, how it predicts novel phenomena, and how it resentational purposes in language, memory, and
compares to other accounts of social information pro-
cessing (also see Barsalou et al., 2003). thought. Damasio (1989) referred to these association
areas as convergence zones. Like feature maps, CZs are
organized hierarchically such that the CZs located in a
Overview of the PSS Theory particular modality-specific system (e.g., vision) cap-
ture patterns of activation in that modality. In turn,
According to PSS, the modality-specific states that higher-level CZs conjoin patterns of activation across
represent perception, action, and introspection in on- modalities. What this means is that when we hear a
line situations are also used to represent these situa- sound (e.g., a fire cracker exploding), conjunctive neu-
tions in the offline processing that underlies memory, rons in auditory CZs capture the pattern of activation in
language, and thought. Rather than using amodal auditory feature maps. Other conjunctive neurons in
redescriptions of online modality-specific states to rep- motor CZs capture the pattern of activation caused by
resent these situations, the cognitive system uses jumping away from the location of the sudden sound.
reenactments (simulations) of them instead. Thus, the And at a higher level of associative processing, con-
key notion in PSS is that simulations of perceptual, junctive neurons in modality-specific CZs conjoin the
motor, and introspective experience underlie the repre- two sets of modality-specific conjunctive neurons for
sentation and processing of knowledge. In the follow- the combined processing of sound and movement.
ing sections, we describe what we mean by modal- It is worth highlighting how the CZ architecture dif-
ity-specific representations in further detail and then fers from traditional ways of conceptualizing knowl-

edge acquisition and use. First, during knowledge ac- tral constructs: simulators and simulations (Barsalou,
quisition (perception and learning), all relevant 1999). Simulators integrate modality-specific informa-
processing regions participate in knowledge represen- tion across a category's instances. Simulations imple-
tation-there is no single "final" region where all ex- ment specific conceptualizations of a category. We de-
perience is abstracted and integrated together. Higher scribe these constructs in turn.
level CZs capture only conjunctions of lower-level
zones (so that CZs can later coordinate their feature- Simulators. A sizable literature on concepts has
level reactivation)-they do not constitute some form demonstrated that categories possess statistically cor-
of "grand" representation that independently repre- related features (e.g., Chin & Ross, 2002; Rosch &
sents all lower levels of the representational hierarchy. Mervis, 1975). Thus, when different instances of the
Second, during knowledge use (e.g., conceptual pro- same category are encountered over time and space,
cessing and recall), the cognizer activates the multiple they activate similar neural patterns in feature maps
modality-specific regions that encoded the experience, (cf., Cree & McRae, 2003; Farah & McClelland,
rather than, as traditionally assumed, only the "final" 1991). One result of this repeated firing of similar neu-
abstract regions at the end of the processing streams. ral patterns is that similar populations of conjunctive
neurons in CZs respond to these regular patterns
(Damasio, 1989; Simmons & Barsalou, 2003). Similar
Reenactments of Modality- to the notion of abstraction, over time, these groups of
Specific States conjunctive neurons integrate modality-specific fea-
What is important about the CZ architecture is the tures of specific categories across their instances and
idea that conjunctive neurons can later reactivate the across the situations in which they are encountered.
states of processing in each modality and across mo- This repetition establishes a multimodal representation
dalities, without any input from the original stimulus. of the category: a concept.
This mechanism provides a powerful way to imple- PSS refers to these multimodal representations of
ment offline embodiment. The modality-specific pro- categories as simulators (Barsalou, 1999, 2003a). A
cessing that occurred in reaction to a previously en- simulator integrates the modality-specific content of
countered stimulus can be reenacted without the a category across instances and provides the ability to
original stimulus being present. For example, when re- identify items encountered subsequently as instances
trieving the memory of a person's face, conjunctive of the same category. Consider a simulator for the so-
neurons partially reactivate the visual states active cial category, politician. Following exposure to differ-
while perceiving it. Similarly, when retrieving an ac- ent politicians, visual information about how typical
tion, conjunctive neurons partially activate the motor politicians look (i.e., based on their typical age, sex,
states that produced it. Indeed, this reentrant mecha- and role constraints on their dress and their facial
nism is now widely viewed as underlying mental imag- expressions) becomes integrated in the simulator,
ery in working memory (e.g., Farah, 2000; Grezes & along with auditory information for how they typically
Decety, 2001; Kosslyn, 1994). sound when they talk (or scream or grovel), motor pro-
Importantly-and different from typical assump- grams for interacting with them, typical emotional re-
tions about imagery-the CZ architecture and PSS do sponses induced in interactions or exposures to them,
not require the reenactment process to be conscious. and so forth. The consequence is a system distributed
Explicit construction of mental imagery can certainly throughout the brain's feature and association areas
produce compelling reenactments, which are often that essentially represents knowledge of the social cat-
viewed as the process that underlies, for example, egory, politician.
counterfactual simulations. Nevertheless, memory, con- According to PSS, a simulator develops for any as-
ceptualization, comprehension, and reasoning pro- pect of experience attended to repeatedly. Because at-
cesses may rely heavily on unconscious reenactments tention is highly flexible, it can focus on diverse com-
(e.g., Barsalou, 1999, 2003b). Many of the demonstra- ponents of experience, including objects (e.g., chairs),
tions illustrated in the previous sections of this article properties (e.g., red), people (e.g., politicians), mental
may primarily reflect spontaneous and unconscious states (e.g., disgust), motivational states (e.g., hunger),
reenactments (e.g., several phenomena reported by actions (e.g., walking), events (e.g., dinners), settings
Bargh and his colleagues). (e.g., restaurants), relations (e.g., above), and so forth.
Across development, a huge number of simulators de-
Simulators and Simulations velop in long-term memory, each drawing on the rele-
vant set of feature and association areas needed to rep-
The CZ architecture describes how knowledge is resent it. Once this system is in place, it can be used to
distributed across the brain's feature and associa- simulate those aspects of experience for which simula-
tion areas. To explain how the cognitive system ac- tors exist. Furthermore, as discussed later, simulators
tually uses that knowledge, PSS relies on two cen- can combine to construct complex representations that

are componential, relational, and hierarchical. Thus and so forth, a unique simulation results (Barsalou,
PSS is not a theory of holistic images. In contrast to 1987, 1989, 1993, 2003b). The simulation process is
how theories like PSS are often mistakenly viewed, context dependent in that the simulation constructed on
photo-like images of external scenes do not underlie a given occasion is tailored to support situated action
knowledge. Instead, componential bodies of accumu- (Barsalou, 2002, 2003b). Ideally, the current simula-
lated information about the modality-specific compo- tion of a category should provide useful inferences
nents of experience underlie knowledge, where these about specific category members currently being expe-
components can represent either the external environ- rienced (or likely to be experienced), actions that could
ment or the internal states of the agent. be performed on them, mental states that might result,
Computational implementations of simulators have and so forth. Thus, PSS does not view the simulation
not yet been developed. Clearly such development is process as producing a static, generic category repre-
important for many reasons. Nevertheless a variety of sentation. Instead, PSS views simulation as a skill or
computational architectures exist that could potentially competency for representing a particular category flex-
be used to implement them. For example, object-ori- ibly in myriad ways that support successful interac-
ented (as opposed to bit-mapped) drawing programs tions with its members.
contain much of the componential, relational, and hier- Notably, simulations do not implement a single rep-
archical structure that underlies PSS's productive use resentational type, such as exemplars. Instead, sim-
of simulations. Although this architecture has not been ulations can implement a variety of representational
developed as a psychological theory, its functionality types, including exemplars, prototypes, and rules (Bar-
closely resembles many of PSS's conceptual opera- salou, 1999, 2003a). To the extent that a simulation re-
tions. Another relevant computational approach is the enacts the modality-specific states of a particular ex-
interactive neural network architecture, which repre- perience with a category member, it represents an
sents categories as distributed feature profiles across exemplar.6 If a simulation, however, draws on multiple
modality-specific areas (Farah & McClelland, 1991; exemplar memories to produce a simulation that is an
O'Reilly & Munakata, 2000). In these networks, mo- average of them, then it functions more as a prototype.
dality-specific representations are used not only for es- Similarly, various types of rules can be implemented
tablishing conceptual knowledge, but also for perform- when relational structures construe particular regions
ing high-level conceptual operations. The success of of simulations as required for category membership.
these existing architectures offers evidence for the Depending on the information simulated and how it is
mechanistic plausibility of the simulators and simula- interpreted, a wide variety of representational types
tions in PSS. can in principle be implemented. Although computa-
tional accounts of these different representational
Simulations. The use of simulators in conceptual types remain to be developed, this approach, in princi-
processing is called simulation. A given simulator can ple, is capable of implementing them.
produce an infinite number of simulations, namely,
specific representations of the category that the simula- Using simulators and simulations to implement
tor represents. On a given occasion, a subset of the mo- basic conceptual functions. Once a collection of
dality-specific knowledge in the simulator becomes simulators exists in long-term memory, it can imple-
active to represent the category, with this subset vary- ment basic functions that are central to a conceptual
ing widely across simulations. For example, a simula- system: types versus tokens, categorical inference,
tor that represents the social category, my significant productivity, propositions, and abstract concepts. We
other, might be used to simulate love making with a address each briefly in turn (for further detail, see
significant other on one occasion, to simulate fights on Barsalou, 1999, 2003b).
another, to simulate quiet togetherness on another, and In the type-token distinction, type representations
so forth. A simulation can be viewed as the reverse pro- stand for categories (e.g., politicians), whereas token
cess of storing modality-specific information in a sim- representations stand for individual category members
ulator. Whereas learning involves feature map infor- (e.g., Napoleon), and more specifically, for individuals
mation becoming linked together by conjunctive units on particular occasions (e.g., Napoleon at Waterloo).
in CZs, simulation involves later using these conjunc- In PSS, simulators represent types because they ag-
tive units to trigger feature map information. Thus, a gregate modality-specific information across category
simulation, too, is a distributed representation. members. Conversely, simulations represent tokens,
According to PSS, the simulation process is highly namely, specific category members, along with spe-
dynamic and context dependent. It is dynamic in that a
given simulator can, in principle, produce an infinite
6When a simulator represents an exemplar, it only produces a
number of simulations. Depending on the current state partial simulation that is typically relatively sketchy and that may be
of the simulator, the current state of associated simula- distorted by a variety of factors. We do not assume that complete
tors, the current state of broader cognitive processing, veridical records of perception are reproduced.


cific category members on particular occasions. Thus, sent above relations such as above (bird, barn), above
the simulator-simulation distinction in PSS naturally (bird, tree), and so forth. Thus, by combining simula-
implements the classic type-token distinction essential tions hierarchically in simulated relational structures,
for a conceptual system. productivity results.
In categorical inference a category member is per- The most basic form of propositional interpretation
ceived, which activates the category representation. results from applying the type-token distinction to the
In turn, knowledge likely to be true for the category is process of categorization. Essentially, the categoriza-
extended to the category member. On seeing a partic- tion process binds a type for a category to one of its to-
ular dog, for example, category knowledge about kens, thereby establishing a type-token proposition.
dogs becomes active, which might then produce in- On seeing a particular overhead projector, for example,
ferences about the dog being likely to bark, wag its categorizing it as an overhead projector binds category
tail, and so forth. In PSS, such inferences arise as knowledge about overhead projectors to the object.
simulations drawn from the modality-specific content This binding represents the proposition that this partic-
of a simulator. Once the perception of the dog acti- ular object is a member of this particular category.
vates the dog simulator (via similarity between the Note that the proposition could be false. The object
content of the perception and the content of the simu- could actually be a piece of abstract art that someone
lator), the simulator runs a simulation of likely per- has mistakenly categorized as an overhead projector. In
ceptual content that has not yet been experienced. A this case, the proposition that represents the agent's be-
major issue for PSS (and for any theory of knowl- lief is false, not true.
edge) is how the correct inferences are generated. In Most notably, the type-token proposition estab-
general though, PSS can produce categorical infer- lished constitutes one possible interpretation of the ob-
ences via the simulation process, simulating likely ject. To see this, consider the infinite number of interpre-
modality-specific content that has not yet been expe- tations of an actual overhead projector that can be
rienced for the perceived individual. implemented with type-token propositions. The projec-
In productivity, concepts are combined systemati- tor could be interpreted correctly as an overhead projec-
cally to construct complex conceptual representations tor, as an office tool, as an artifact, as an increasingly
(e.g., productively combining striped and purple with dated piece of technology, and so forth. Similarly, the
waterfall and river yields striped waterfall, purple wa- projector could be interpreted incorrectly as a piece of
terfall, striped river, and purple river). Notably, the art, as a mammal, or as a space alien. In each case, differ-
conceptual combination process is capable of repre- ent categorical knowledge is applied to the same object
senting an infinite number of concepts whose referents to create a different type-token proposition. In each
have never been experienced, such as purple waterfall. case, a different interpretation results, accompanied by a
Because PSS establishes simulators for individual different family of categorical inferences.
components of experience, it has the necessary build- It has been widely argued that modality-specific ap-
ing blocks for implementing productivity. Once simu- proaches to knowledge cannot implement proposi-
lators for striped, purple, waterfall, and river exist, they tions, because these approaches implement holistic im-
can be combined to form more complex concepts. For ages that record experience, rather than implementing
example, a person could run the waterfall simulator to concepts that interpret it (e.g., Pylyshyn, 1973). In
produce a particular waterfall simulation. Once this PSS, however, the binding of simulators to perceived
simulation is in place, then the color of the waterfall or simulated individuals naturally implements propo-
can be systematically varied, using simulators for sitions. Because PSS implements the type-token dis-
color, such as purple, orange, and gold, to differentially tinction using simulators and simulations, type-token
simulate the waterfall's color. propositions are a natural consequence.
Central to productivity is having relational knowl- Finally, it is often argued that modality-specific ap-
edge about how various components combine to form proaches fail because they cannot represent abstract
more complex structures. Just as PSS establishes simu- concepts such as truth. Furthermore, it is often as-
lators for objects and properties, however, it also estab- sumed that these approaches fail because representa-
lishes simulators for a wide variety of relations, such as tions of the external world cannot be used to represent
the aforementioned. To productively simulate different abstract ideas. As we have seen, however, PSS estab-
above relations, PSS first uses the simulator to con- lishes simulators, not only for components of the ex-
struct a configuration of two regions that were previ- ternal world, but also for components of introspection,
ously established as one instance of this relation (e.g., including emotions, motivational states, cognitive op-
two spherical regions equal in horizontal position, dif- erations, and so forth. Barsalou (1999) proposes that
ferent in vertical position, nearly touching each other). abstract concepts are abstract because they focus
Once this relational simulation is in place, its regions heavily on introspections and complex situational
can be systemically filled with simulations of different events. In contrast, concrete concepts are concrete be-
objects, such as birds, planes, barns, and trees, to repre- cause they focus on physical entities, settings, and sim-

ple behaviors in the external world. Because simulators likely to occur while interacting with it. For example, if
can be established for introspections and events (not the conceptualizer were on a jet, a chair simulation
just for concrete objects), they can in principle repre- would take the form of a jet chair, along with the rele-
sent the conceptual content of abstract concepts (not vant actions and mental states for interacting with it ef-
just the content of concrete concepts). fectively. Conversely, if the conceptualizer were in a
To assess this hypothesis, Barsalou and Wiemer- movie theater, a chair simulation would take the form
Hastings (in press) used the property listing task to as- of a theater chair, along with the appropriate actions
sess the content of abstract concepts (truth, freedom, and mental states. Thus, a situated conceptualization
invention) and of concrete concepts (car, sofa, bird). simulates the focal category entity, along with simula-
After participants listed the properties of these con- tions of the likely setting, actions, and introspections.
cepts, detailed coding schemes were applied to assess Because the simulation includes the conceptualizer's
the content produced. Most notably, the general types actions and introspections, the simulation creates the
of content for abstract and concrete concepts were experience of "being there" with the category member.
highly similar. For all concepts, participants tended to As a result, the conceptualizer is well prepared to inter-
describe situations that included objects, people, set- act with the entity in the anticipated situation.
tings, behaviors, events, mental states, and relations. Barsalou et al. (2003) propose that the construct
For both types of concepts, participants situated their of situated conceptualization is useful in explaining
conceptualizations of them, not just representing the the wide variety of embodiment effects reported in
focal category content, but also representing extensive social psychology. Specifically, they propose that sit-
background situational content relevant to understand- uated conceptualizations for repeated social situations
ing and using the category. become entrenched in memory. Consider, for example,
The two types of concepts differed in their focus on the repeated situation of parents dealing with an upset
this content. Whereas concrete concepts focused on child. Across these repeated situations, a situated con-
entities, settings, and simple behaviors, abstract con- ceptualization becomes entrenched that represents
cepts focused on introspections, social entities, and how the child typically appears, how the parents feel,
complex events. Furthermore, the abstract concepts what the parents do, and so forth. Because these sit-
were more complex, including greater relational struc- uations contain many embodiments (e.g., perceptual
tures, organized in greater hierarchical depth. simulations of events, bodily states associated with
This exploratory study did not assess how partici- emotion and action), these embodiments become rep-
pants represented this content. In principle, though, it resented in the situated conceptualization. Later, these
seems possible that all of this content-for both con- embodiments, when experienced, can trigger the situ-
crete and abstract concepts-could be simulated. Ev- ated conceptualization via the inference process of pat-
erything that participants mentioned is something ex- tern completion. Specifically, the experienced embodi-
perienced either in the external or internal world. As a ment activates a larger pattern that contains it, with
result, the process that establishes simulators could act nonperceived aspects of the pattern constituting infer-
on this content and establish simulators for it, thereby ences about the situation. Conversely, if through lin-
making it possible for a theory like PSS to explain it. guistic conversation, the situated conceptualization be-
Notably, amodal theories face the same problem of comes active, it can, in turn, produce corresponding
specifying the content of abstract concepts and of ex- embodiments via the same inference process.
plaining how their representations implement this con- The PSS framework, with its construct of mul-
tent, something that is far from having been accom- timodal, situated conceptualization, accounts for the
plished satisfactorily. diverse collection of social embodiment effects re-
ported in the literature. Consider priming effects on be-
havior, as when exposure to words associated with the
Situated Conceptualizations
elderly stereotype produces slower walking (Bargh et
As we just saw, people situate their representations al., 1996). On the PSS account, stereotypes are situated
of categories. Thus, situated conceptualizations consti- multimodal conceptualizations of social categories. In
tute another central construct in the PSS framework. A the case of the elderly stereotype, its content includes
situated conceptualization is one particular simulation embodiments of slow motor movements. Activating
of a category in a background situation, where the spe- these embodiments during conceptualization of the
cific content of the category and situation prepare the stereotype influences action, via top-down processing,
conceptualizer for action in it (Barsalou, 2002, 2003b). in related modality-specific systems, as when walking
In representing the category of chairs, for example, a slowly toward the elevator. In another example, con-
conceptualizer does not just simulate a generic repre- sider how bodily movements influence conceptual pro-
sentation of chairs in a vacuum. Instead, the concept- cessing, as when head nodding during a persuasive
ualizer simulates one particular kind of chair in a par- message leads to more positive attitudes (Wells &
ticular setting, along with actions and mental states Petty, 1980). On the PSS account, understanding a

message produces a multimodal situated conceptual- the property to assess whether the property actually
ization to represent the messages' meaning. Nodding belonged to the concept. Consistent with this predic-
while representing the message's meaning activates tion, Solomon and Barsalou (2004) found that, under
multimodal conceptualizations for prior situations in these task conditions, the associative strength between
which positive affect occurred. Once positive affect be- concept and property words best predicted verification
comes active, it influences both how the message is times and errors.
represented and how it is evaluated, producing greater When, however, the false properties were always re-
positivity than if no action or a negative action were lated to their concepts, participants could not rely on
performed instead. superficial associations between the concept and prop-
erty words, given that the false trials possessed them
too, not just the true trials. As a result, participants had
Shallow Versus Deep Processing to access conceptual knowledge to verify that the con-
Another important construct in the PSS framework cepts indeed possessed the properties. Solomon and
is the distinction between shallow versus deep process- Barsalou (2004) predicted that if simulators represent
ing. PSS does not require that all cognitive tasks utilize conceptual knowledge, then perceptual variables
simulation. Following Paivio (1986) and Glaser should become the best predictors of verification time
(1992), PSS assumes that people can also use word- and errors under these task conditions-not the asso-
level representations to perform superficial "concep- ciative strength between concept and property words.
tual" processing when task conditions permit. Con- In support of this prediction, perceptual variables, such
versely, when task conditions block superficial word as property size, became the best predictors of perfor-
strategies and force people to perform conceptual pro- mance. When participants were forced to abandon the
cessing, simulations come into play. Not surprisingly, superficial word association strategy, the conceptual
participants adapt flexibly to task conditions. When knowledge they used appeared to take the form of per-
they can use shallow processing, they do, but when ceptual simulation.
they cannot, they perform deep processing. To corroborate this conclusion, Kan et al. (in press)
To see how participants adapt flexibly to task condi- ran the same experiment in an fMRI scanner. When
tions, consider an experiment by Solomon and Bar- participants received only related false properties, vi-
salou (2004; also see Niedenthal et al., 2002). On each sual processing areas in the fusiform gyrus became
trial, participants read the word for a concept (e.g., active to simulate the properties conceptually. When
"CHAIR"), and then verified whether a subsequently participants received only unrelated false properties,
presented property (e.g., "seat") was true of the con- however, these areas were no longer active, supporting
cept (where the correct response on true trials, such the conclusion that participants were using the superfi-
as this one, was "yes"). The key manipulation was cial word association strategy instead, given that task
whether the properties tested on false trials (where the conditions allowed. As all these experiments illustrate,
correct response was "no") were completely unrelated conceptual processing is flexible, and participants
to the target concept (e.g., CHAIR-feathers), or were need not always use simulation. When conditions al-
associatively related to it (e.g., CHAIR-table; note that low, participants adopt more superficial strategies.
a true property had to be a part of the target concept).
Whereas some participants only received unrelated
false properties, others received only related false Support for the Simulation Hypothesis
properties, with both groups verifying the same true
properties. Increasing evidence supports the central assump-
Solomon and Barsalou (2004) predicted that the re- tion of PSS that simulation underlies conceptual pro-
latedness of the false properties would determine cessing. Here we focus on several lines of supporting
whether participants used a shallow or deep strategy evidence reported by Barsalou and his colleagues, al-
for verifying the true properties. When the false prop- though much evidence has been accumulating in other
erties were completely unrelated (e.g., CHAIR-feath- laboratories as well (for reviews, see Barsalou, 2003b;
ers), participants could employ a shallow processing Hegerty, 2004; Martin, 2001). Il particular, we focus
strategy. Because all the true properties were related to on three types of evidence for simulation: modality
their respective concepts, whereas all the false proper- switching costs, instructional equivalence, and percep-
ties were unrelated, simply detecting an association be- tual effort.
tween the concept and property words was adequate
for determining the correct response. Whenever an as- Modality Switching Costs
sociation was detected, participants could correctly re-
spond true; whenever they did not detect one, they PSS predicts that switching costs should occur
could correctly respond false. Participants did not need when people verify properties on different modalities.
to access conceptual knowledge about the concept and Pecher, Zeelenberg, and Barsalou (2003, 2004) tested

this prediction in a series of studies using the property PSS predicts that participants in the imagery and
verification task (also see Marques, 2004). Participants concept conditions should produce similar profiles of
were first asked to verify a property of a concept that properties. If concept participants produce simulations
requires simulation in one modality, such as BLEND- by default, the representations that they use to produce
ER-loud (which requires an auditory simulation). properties should be highly similar to those that imag-
Next, participants were asked to verify a property of a ery participants produce. Furthermore, the profiles in
second concept, which either required simulation in the concept and imagery conditions should differ from
the same modality, such as LEAVES-rustling (again those in the word association condition. Because word
requiring an auditory simulation), or in different mo- association participants rely primarily on the associa-
dality, such as LEMON-sour (requiring a gustatory tive lexical system, they minimize use of the concep-
simulation). tual system, adopting a shallow processing strategy
As PSS predicts, switching costs occurred. When that produces the requested responses.
the modality of the property changed from the first trial Amodal theories most naturally make a different a
to the second, participants were slower to verify the priori prediction (e.g., feature list and semantic net-
second property than when the modality stayed the work models). According to these theories, concept
same. This finding suggests that when people represent participants should, by default, use amodal represen-
properties, they simulate them in the respective modal- tations, not simulations. As a result, their property pro-
ity-specific systems. Amodal theories of knowledge do files should differ significantly from the property pro-
not predict switching costs a priori. Instead, these theo- files of imagery participants. Furthermore, some
ries assume that properties are represented amodally amodal approaches, at least, predict that the profiles of
and can be verified without activating modality-spe- concept participants should be similar to those of word
cific systems. For a review of modality switching ef- association participants. Because amodal theories of-
fects and for further discussion of their theoretical in- ten assume that symbols for concepts bear a systematic
terpretation, see Barsalou, Pecher, Zeelenberg, relation to the units of language, they lend themselves
Simmons, and Hamann (in press). to the prediction that the default strategy of concept
participants is to activate symbols that correspond to
the words produced in the word association condition.
Findings from Wu and Barsalou (2004) and Krauth-
Instructional Equivalence Gruber et al. (2004) support the PSS predictions. Across
PSS predicts that people, by default, engage in per- multiple experiments, the property profiles for the con-
ceptual simulation. One test of this prediction involves cept and imagery conditions were more highly corre-
examining instructional equivalence, or whether par- lated with each other than with the profiles for the word
ticipants perform similarly under instruction condi- association condition. Furthermore, the correlations
tions that emphasize or do not emphasize simulation between the concept and imagery condition were high,
(Barsalou, Solomon, & Wu, 1999). Recent research suggesting that concept participants adopted the same
has recently used the property verification task to as- representational strategy used in the imagery condition
sess instructional equivalence (Krauth-Gruber, Ric, & by default (i.e., simulation).
Niedenthal, 2004; Wu & Barsalou, 2004). Across stud-
ies, participants in the imagery condition were explic- Perceptual Effort
itly asked to image a referent of a concept before listing
its properties (e.g., construct an image of a chair and Perceptual effort constitutes another indicator of
then describe properties in the image). Much previous simulation by default (Barsalou et al., 1999). If people
work indicates that imagery instructions typically in- spontaneously use simulations to represent concepts,
duce participants to construct images, which are typi- then these representations should have perceptual
cally heavily visual. In contrast, participants in the qualities. As a result, manipulating perceptual vari-
concept condition received no special instructions and ables should affect the ease of conceptual processing,
were simply asked to produce properties that are typi- much like manipulating these variables affects the ease
cally true of the concept. The primary prediction from of processing of mental images in working memory
PSS is that concept participants will spontaneously, by (e.g., Finke, 1989; Kosslyn, 1980; Shepard & Cooper,
default, construct simulations, much like those con- 1982). Manipulating perceptual variables such as size,
structed by imagery participants, although perhaps less orientation, and occlusion should influence the ease of
vivid and less conscious. Finally, participants in the processing concepts and their properties. As simula-
word association condition were asked to generate tions of a property become increasingly large, for ex-
words associated with the concept name. In each in- ample, greater time and effort are needed to construct
structional condition, participants' protocols were it. Similarly, the more an object simulation must be ro-
coded using a detailed coding scheme that established tated to achieve an upright position, the greater the
a profile of the conceptual content produced. time and effort needed to complete this operation.

Wu and Barsalou (2004) reasoned that if people and expectancy variables. Under conditions that re-
are representing object properties with simulations, the quired conceptual processing (i.e., related false
perceptual variable of occlusion should affect the ease properties as described earlier), the perceptual vari-
generating properties in the property generation task. ables predicted performance better than the linguistic
To see this, imagine participants being asked to pro- and expectancy variables. In particular, the size of
duce the properties of, say, watermelons. If perceptual properties was important in predicting performance.
effort affects this task, then participants should pro- As properties became larger, they took increasingly
duce the properties that require the least effort to per- longer to verify and led to higher rates, presumably be-
ceive in their simulations, namely, those that are cause large properties take more time and effort to rep-
unoccluded. Producing occluded properties should re- resent than small ones. Under conditions that did not
quire more perceptual effort, because simulation must require conceptual processing (i.e., unrelated false
be transformed to reveal them. Thus, for watermelons, properties), the linguistic variables best predicted per-
participants should produce outer unoccluded proper- formance, suggesting that participants used the super-
ties, such as green and stripes more often than occluded ficial word association strategy instead of deeper con-
properties such as red and seeds. Across several experi- ceptual processing, as described earlier.
ments, Wu and Barsalou found evidence for this pre-
diction. Unoccluded properties were produced more
often than occluded properties, and also earlier and in Related Views
larger clusters.
Wu and Barsalou (2004) also tested a second pre- In this article, we propose that the PSS account is
diction. If the normally occluded properties of an ob- productive for understanding embodiment phenomena
ject become unoccluded, they should require less per- in social psychology. Previous theories in social psy-
ceptual effort to produce and therefore be produced chology, however, have made related proposals. Thus,
more often. Thus, in some conditions of these experi- it is useful to compare and contrast PSS with these re-
ments, participants produced properties of the same lated accounts. The two accounts most relevant are the
objects but where the name of each object included a Hard Interface Theory (Zajonc & Markus, 1984) and
revealing modifier. Rather than producing the proper- the Associated-Systems Theory (Carlston, 1994).
ties of watermelons, for example, these participants
were asked to produce the properties of a half water- Hard Interface Theory (HIT)
melon. Because the normally occluded inner parts of a
watermelon become unoccluded in a half watermelon, In 1984, Zajonc and Markus published an influen-
less effort is required to perceive them in a simulation. tial chapter titled "Affect and cognition: The hard inter-
Thus, normally occluded properties should increase in face." In this chapter, Zajonc and Markus encouraged
production rates, becoming more comparable to nor- social psychologists to pay more attention to embodi-
mally unoccluded properties. Across several studies, ment, and in this spirit they introduced the HIT. Al-
Wu and Barsalou (2004) observed this pattern, consis- though, Zajonc and Markus focused primarily on the
tent with the prediction that participants were using interaction between emotion and cognition, they also
simulations to represent the concepts. made general proposals about the representation of so-
Krauth et al. (in preparation) found that producing cial knowledge, which can be viewed as precursors to
the properties of an emotional state produced a similar more recent developments. In particular, Zajonc and
occlusion effect. When participants were asked to de- Markus (1984) declared that purely propositional,
scribe properties of "his anger," they produced oc- "disembodied" theories of social knowledge are unsat-
cluded properties of the other's mental state less often isfactory for several reasons.7 For one, propositional
than when they produced properties of "my anger," theories rarely address how abstract representations
when these mental state properties are unoccluded. are connected to actual behavior. Thus, in the emotion
Again, the amount of effort required to simulate prop- domain, there is no clear account of how cognition
erties determined their rates of production. (e.g., "hearing an insult") elicits bodily components of
To further explore perceptual effort, Solomon and anger (clenched fists, red faces, bulging veins, etc). A
Barsalou (2004) again employed the property verifi- related problem arises when one tries to understand the
cation task (also see Solomon & Barsalou, 2001). If influence of emotion on cognition. In propositional
people simulate properties to verify them in their re- theories, the problem is reduced to the influence of one
spective objects (e.g., PONY-mane), then perceptual associative structure on another by assuming that the
variables should explain the time to verify properties,
and also the accompanying error rates. To assess this 7The term propositional is used widely throughout cognitive and
issue, Solomon and Barsalou regressed verification social psychology when referring to amodal theories of knowledge,
RTs and error rates onto potential factors that could even though, as we have seen, embodied theories can also be propo-
predict performance, including linguistic, perceptual, sitional (Barsalou, 1999, 2003a).


bodily (muscular, hormonal) components of emotion people's particularly good memory for faces reflects
somehow get transduced into a propositional form. their ability to imitate perceived faces and to cre-
Finally, propositional theories have a hard time ac- ate hard muscular representations that complement
counting for embodiment phenomena in cognition. soft representations. In a similar fashion, Zajonc and
Why, for example, Zajonc and Markus asked, "do peo- Markus proposed that mood congruence effects occur
ple engaged in an arithmetic problem often gnash their because affective states create bodily configurations
teeth, bite their pencils, scratch their heads, furrow that facilitate the production of compatible responses.
their brows or lick their lips?" (p. 74), "Why do people The state of happiness, for example, creates a specific
scratch their heads and rub their chins when they try to posture and muscle tone, which facilitates "happy"
remember something?" (p. 84). muscular responses to objects. In yet another example,
Zajonc and Markus's (1984) response is that bodily Zajonc and Markus suggested that the mere-exposure
states constitute "hard representations." According to effect results from the gradual relaxation of bodily re-
this view, bodily states have representational content. sponses to a stimulus as a result of habituation, thereby
When a dog hears a bell, for example, and withdraws a making positive evaluation more compatible with the
leg to avoid shock, this leg flexion can be said to repre- bodily state.
sent the dog's knowledge of the bell-shock relation. Despite the shared affinity for embodiment, differ-
Importantly, HIT proposes that the bodily movement ences exist between HIT and PSS. Perhaps most sig-
itself has representational content and does not require nificantly, HIT assigns representational functions to
a more cognitive "soft representation" to have repre- actual bodily states. This includes not only muscles,
sentational value. A purely hard representation, such as but also gastrointestinal, glandular, and cardiovascu-
the dog's leg flexion, can guide behavior in a danger- lar systems. In the original PSS account (Barsalou,
ous situation as effectively as an abstract soft represen- 1999), the critical mechanism is simulation in the
tation of the event. Individual hard representations can brain, where the body plays a role only as represented
also interact with one another, as when, for example, in neural systems. A related difference is that HIT al-
muscles associated with a forward movement (hard lows representational functions of bodily states to
representation of positivity) interfere with muscles as- work "by themselves," without translation into "soft"
sociated with a backward movement (hard representa- representations. In contrast, the original PSS account
tion of negativity). required that bodily states are first encoded into neu-
Probably the best known empirical illustration of ral representations in modality-specific systems,
HIT is the "chewing gum" experiment (Zajonc et al., which then allow bodily states to interact with cogni-
1982; cf. Graziano, Smith, Tassinary, Sun, & Pilking- tion. In subsequent developments of PSS, however,
ton, 1996). Participants were first asked to study 78 bodily states have been central to the situated concep-
photographs of faces in four between-participant con- tualizations that underlie higher cognition-not just
ditions. One group imitated the targets' head and gaze simulations of these states (e.g., Barsalou, 2003b;
orientations and their facial expressions (instructed Barsalou et al., 2003).
mimicry). A second group chewed gum (blocked mim- Another difference in relative emphasis between the
icry). A third group squeezed a sponge with their two theories concerns associative versus dynamic pro-
nonpreferred hand (motor control). A fourth group cessing. HIT is fundamentally associative, establishing
judged the head orientations and facial expressions in relatively stable associations between particular cog-
the photographs (judgment control). After studying the nitions and their accompanying embodiments. PSS
pictures, participants received a recognition test. As contains many associative mechanisms as well, but fo-
the embodiment view predicts, memory was best when cuses more on the dynamic construction of situational
participants' embodiments were compatible with the conceptualizations and on the productive construction
pictures-the instructed mimicry group scored highest of simulations into complex relational structures that
(73% correct). The worst performance occurred for function prepositionally. Analogously, HIT has focused
participants who performed the most competitive mo- more on explaining social phenomena, whereas PSS
tor response-chewing gum (59%). Participants who has focused more on explaining a variety of higher
squeezed a sponge fell in between (65%) as did partici- cognitive functions, such as type-token processing in
pants who judged the faces (64%). This pattern of re- categorization and the productive construction of com-
sults suggests that participants represented properties plex simulations in language comprehension.
of the perceived faces in their musculature, such that
when their musculature was consistent with a face, it
enhanced memory via consistent elaboration. Associated-Systems Theory (AST)
Another important aspect of HIT is that it provides AST provides a multimodal model of how the cog-
intriguing accounts of classic psychological phenom- nitive system represents other people in person percep-
ena that differ considerably from standard accounts. tion (Carlston, 1994). According to AST, representa-
For example, Zajonc and Markus (1984) proposed that tions of other people develop through the use of four

primary mental systems: (a) the visual system, (b) the judgment but did not alter the influence of verbal infor-
verbal/semantic system, (c) the affective system, and mation. Conversely, Claypool and Carlston found that
(d) the action system. Each system captures the rele- interference in the verbal system decreased the influ-
vant features of external stimuli and produces repre- ence of verbal information but not the influence of ap-
sentations that are specific to it. Thus, the visual sys- pearance information. Depending on the information
tem represents a target person's appearance, the action presented and the availability of relevant systems, dif-
system represents an action sequence directed at the ferent profiles of processing for a target ensued.
target person, the affective system represents affect to AST shares many assumptions with PSS, in particu-
the target, and the verbal system represents the target's lar, the central assumption that multiple modality spe-
personality traits. cific systems underlie cognitive processing. Both theo-
Each system in AST is hierarchically organized. ries also view CZ theory as an appropriate description
Whereas the lowest levels contain highly specialized of knowledge representation in the brain (Damasio,
structures used to perceive stimuli and produce re- 1989). Finally, AST and PSS both assume that process-
sponses, the highest levels contain abstract concepts ing is dynamic. Depending on current task conditions,
related associatively to perceptual or response pro- different representations and different forms of pro-
cesses. At the lowest levels, the four primary systems cessing control task performance.
are independent of each other. At higher levels, how- AST and PSS differ, however, on other points. First,
ever, they become progressively intertwined, interact- AST gives primacy to the abstract verbal system, with
ing to produce hybrid forms of representation. When its representational elements of words and personality
categorizing individuals into social groups, for exam- traits. According to PSS, however, even abstract con-
ple, a perceiver accesses both an image of the group (in cepts like personality traits are grounded in modal-
the visual system) and descriptive labels for their traits ity-specific systems, as are the representation of words
(from the verbal system). Similarly, when evaluating a in the respective modalities (e.g., orthographic repre-
target person, a perceiver uses both the verbal and the sentations of the word in the visual system and acoustic
affective systems. Categorizations and evaluations, to- representations of words in the auditory system; Bars-
gether with orientations (a hybrid of affect and action alou, 1999; Simmons & Barsalou, 2003). Thus, PSS is
system) and behavioral observations (a hybrid of vi- somewhat more radical than AST in its reliance on the
sual and action systems), constitute the four products modalities as the basis of representation, whatever its
of AST's secondary systems. degree of abstractness.
According to AST, the representations that one es- Second, AST, as its name implies, attempts to ex-
tablishes for a target person typically utilize one distri- plain the role of modal information in terms of associa-
bution of the four primary systems more than other tive processes. Accordingly, the previously mentioned
possible distributions. For instance, passive observa- limitations of HIT also apply to AST. In particular, it is
tion of a target person produces appearance representa- not clear how AST accounts for classic conceptual
tions grounded mainly in the visual system. In contrast, functions such as the type-token distinction, produc-
actually interacting with a person activates not only the tivity, propositional interpretation, and the representa-
visual system, but also the action system, thereby pro- tion of abstract concepts.
ducing a different distribution. In a related manner, re-
sponses to a target person depend on the distribution of
system(s) solicited in the response situation. Criticisms of Embodiment Theories
Furthermore, because AST's systems can function
independently of each other, the representations in dif- Theories of embodied cognition have appeared reg-
ferent systems can potentially conflict when processing ularly in the cognitive and social literatures. So far,
a given individual. When such contradictions occur, however, these theories have not become a major force
judgments of the target depend on which of the conflict- in explaining cognition, in general, and social cogni-
ing systems dominates the judgment situation. If the vi- tion, in particular. The major reason is previous ac-
sual system dominates (e.g., because the target is pre- counts have had trouble dealing with certain classes of
sented via a picture), responses to the target will be criticism. In the preceding sections, we have already
based on appearance, even when contradictory judg- mentioned how PSS goes beyond earlier proposals in
ments arise in other systems. If, however, the verbal sys- social psychology. Here we list criticisms that have
tem dominates (e.g., because the target is presented via been raised against embodiment theories and show
text description), responses to the target will be based on how PSS addresses them.
personality traits. To demonstrate the relative independ-
ence of the four systems and the potential for one to Selective Embodiment
dominate others, Claypool and Carlston (2002) showed
that interference in the visual system at encoding de- Embodiment theories need to solve the problem of
creased the influence of visual information in liking how some cognitive activity can proceed without in-

volvement of bodily states and modality-specific simu- problem that the same bodily state may be associated
lations. PSS addresses this issue with the distinction with different cognitive and emotional representations
between shallow versus deep processing. A perceiver (Zajonc & McIntosh, 1992). These issues are actually
simulates primarily when needed. When conceptual quite old and have been effectively used by Cannon
tasks can be solved using shallow strategies, such as and many others to argue against the James-Lange the-
word association, simulations of conceptual content ory of emotion (James, 1896/1994).
are not recruited or play only peripheral roles. Several responses to this criticism are possible.
Zajonc and Markus (1984) note that the motor system
can support extremely subtle distinctions, as the so-
Dynamic Use of Embodiment
phistication of sensory-motor processing in spoken
A related challenge to embodiment theories arises language illustrates. Further, even a limited number of
from findings suggesting that perceivers are quite flexi- bodily states can support a very large number of repre-
ble in their use of embodied information. For example, sentational distinctions. (Consider how many melodies
when empathizing with others, people typically rely on can be played with 88 piano keys a number much
their own modality-specific reactions, suggesting use of lower than the number of muscles in the body.)
a simulation strategy (Decety & Chaminade, 2003; More important, recent embodiment approaches,
Gallese, 2003). However, the exact nature of the stimu- such as PSS, CZ theory, and Somatic Marker Theory
lation depends on task goals. For example, in some (Damasio, 1999), avoid the "body-is-too-crude-too-
cases, people empathize by simulating emotional states, slow-and-too-varied" criticisms by focusing on the
whereas in other cases people empathize by simulating brain's modality-specific systems, instead of on actual
cognitive states. In some situations, people may decide muscles and viscera. The circuits in modality-specific
not to simulate at all (e.g., when dealing with a mass brain areas are as fast and refined as any other form of
murderer). More generally, according to modern em- cortical representation and are thus able to flexibly pro-
bodiment theories such as the PSS, whether or not peo- cess a large number of modal states at the same time
ple decide to simulate, what specific modality people (Damasio, 2003).
decide to simulate in, and how they are going to use the
products of the simulation is open to task goals.
The notion of goal-driven stimulation allows mod- Higher Cognitive Functions
ern embodiment theories to account for social psy- Embodiment theories must account for the basic
chological findings showing people's flexibility in us- functions of higher cognition, such as the type-token
ing experiential sources of information in judgment. distinction, categorical inference, productivity, propo-
Thus, people typically draw on experiential feedback sitional interpretation, and abstract concepts. As dis-
(mood, feelings of processing difficulty, etc.) when cussed earlier, PSS uses the constructs of simulators
solving a variety of cognitive tasks, including fre- and simulations to explain these phenomena. Because
quency judgments, self-assessment, and value judg- simulators develop for components of experience, cat-
ments (see Schwarz, Bless, Waenke, & Winkielman, egorical knowledge results that represents types, pro-
2003). When people realize that their experiential reac- duces categorical inference, and combines produc-
tions are not diagnostic, however, they switch to alter- tively. Furthermore, simulators become bound to both
native sources ofjudgments, making less use of current perceived and simulated individuals to implement
experience. Critics could interpret this finding as sug- type-token propositions. Because PSS uses selective
gesting that people use simulation only under default attention to break experience into components and then
conditions, and that under conditions that produce establish categorical knowledge about them, it devel-
misattributions, people fall back on amodal process- ops the classic capabilities of a conceptual system.
ing. Instead, we propose that such findings reveal peo- Perhaps the most common criticism of theories of
ple's flexibility in their use of the simulation strategy. embodied cognition concerns their ability to represent
When an external factor compromises validity of a abstract concepts. An implicit assumption of these crit-
simulation in one modality, people can switch to a sim- icisms, however, is that only simulations of external
ulation in an alternative modality. experience can be used for representational purposes in
embodiment theories. As we have seen throughout this
Representational Limitations article, however, simulations of introspective experi-
of the Body ence are also available for representational purposes.
Furthermore, as we saw earlier, simulations of intro-
Embodiment theories must address the problem of spections are central to the representation of abstract
the body's representational capacity. Starting with concepts (Barsalou & Wiemer-Hastings, in press).
Cannon (1927, 1929), critics have argued that bodily More generally, the simulation of situations appears to
feedback is too undifferentiated and too slow to serve provide a natural approach to representing a wide vari-
as the basis of experience. Furthermore, there is the ety of concepts, including abstract ones.

Regressing to Behaviorism can mimic any observed (or even never-observed!)

or Mere Associationism? finding (e.g., Anderson, 1978; Pylyshyn, 1981). This
makes these models eminently powerful but also un-
Closely related to the issue of higher cognitive func- falsifiable. Because these models can explain anything,
tions is the worry that embodiment theories advocate a
it is impossible for any given result to disconfirm them
return to the view that simple associations between
entirely. Thus, simply saying that semantic memory
sensory and motor processes underlie all cognition.
models, or any other amodal model, can account for
Long ago, behaviorist theories were conclusively
shown to be incapable of explaining cognitive achieve-
embodiment effects is not very compelling.
ments such as memory, language, and thought (Chom-
Second, the older and entrenched status of amodal
theories does not necessarily endow greater credibility.
sky, 1959). We endorse these conclusions strongly and
What would give them more credibility is the ability to
agree that simple associations between sensory-motor
predict embodiment effects a priori. Notably, however,
states, drawn only from experience, cannot explain
amodal theories were not developed with embodiment
cognitive processes. What we propose here shows how
effects in mind and thus have not been used to predict
the sophisticated symbolic processing characteristic of
and explain such effects. For example, even though as-
human cognition is grounded in embodied representa-
sociative models of emotion memory can in principle
tions. An associationist account is as inadequate now
as it was 50 years ago.
explain embodiment effects in emotion, such models,
to our knowledge, have never motivated researchers to
predict embodiment effects in an experiment. If any-
Are Observed Mind-Body thing, in fact, such models have been repeatedly criti-
Relationships Epiphenomenal? cized for their implausible use of informational units to
stand for expressive and autonomic processes and for
Embodiment theories make clear claims of causal- their somewhat vague declaration of interactions be-
ity (e.g., "modality-specific systems areas are critical tween those types of units and representations of more
for conceptual processing"). Nevertheless, supporting abstract information (Niedenthal, Setterlund, & Jones,
evidence has sometimes been viewed as only correla- 1994). Just because amodal theories can be configured
tional in nature, consisting mainly of observations post hoc to explain any embodiment effect is not im-
of co-occurrence between embodiment and cognition. pressive. What would be more impressive is if they pre-
Thus, critics have argued that perceptual representa- dicted these effects a priori. Again, to our knowledge,
tions are not constitutive of concepts but only become they never have.
active epiphenomenally during the processing of Finally, unless one is very impressionistic in his or
amodal symbols, which themselves do the work of her reading of amodal theories, embodiment effects
high-order cognitive processing. As reviewed through- simply do not follow naturally from them. At the heart
out this article, however, theories of embodied cogni- of these theories lie two assumptions of modularity:
tion can now draw on many experimental findings that (a) the conceptual system is functionally and physi-
show that manipulations of embodiment causally mod- cally separate from modality-specific systems, and (b)
ulate cognitive and social performance. Demonstrating knowledge abstracts over the details of experience. For
that the inhibition or facilitation of a specific motor these reasons, we believe that embodiment effects vio-
behavior, or a specific modality-specific resource, cor- late the very spirit of amodal theories. Although there
respondingly inhibits or facilitates conceptual process- may not be a 'killer" empirical finding that the amodal
ing indicates relations between embodiment and cog- approach cannot explain (to the extent that the whole
nition that are not merely epiphenomenal. framework can ever be challenged by an empirical
finding), it is becoming increasing clear, at a minimum,
that the embodiment approach offers great value in
Amodal Theories Make
the Same Predictions
pointing out new and exciting research directions.
One primary criticism of embodiment theories is
that the empirical evidence for them can be explained
by amodal theories. For example, critics often suggest Explaining Embodiment Effects
that semantic network models (Collins & Quillian, in Social Psychology
1969) and emotional memory models (Bower, 1981)
can account for embodiment effects. Although this is The prevailing theoretical approaches in social psy-
technically right, such an objection is being taken to chology view modality-specific and bodily states as
task (e.g., Barsalou, 1999; Glenberg, 1997; Lakoff, peripheral to social knowledge. The findings reviewed
1997; Newton, 1996). here, however, indicate that these states play a more
One problem for these accounts is their uncon- important role in information processing than previ-
strained expressive power. In theory, amodal theories ously imagined. Here we come full circle and suggest

how theories of embodied cognition contribute to the Mood Congruence

understanding of social phenomena.
One of the most studied effects in social psychology
is mood congruence-the influence of previously in-
Stereotypes and Traits duced emotion or mood on subsequent judgments. The
literature, however, is replete with inconsistent results
Many researchers assume that distilled amodal rep- and accounts (Forgas, 1995). Embodied accounts can
resentations constitute the core conceptual representa- explain why and when emotion congruence occurs.
tions of stereotypes and traits, with embodiments being This has useful implications for understanding a wide
irrelevant. Other researchers assume that embodiments range of specific congruency effects, including the in-
have some relevance for social processing but that em- fluence of mood on judgments (Bless & Fiedler, 1995;
bodied states are associated peripherally with traits and Schwarz & Winkielman, 1999), affective priming
stereotypes, rather than constituting their core concep- (Niedenthal, 1990; Winkielman, Zajonc, & Schwarz,
tual content. In contrast to both views, embodiment 1997), and affect and memory (Eich & Macaulay,
theories hold that multimodal simulations of percep- 2000).
tion, action, and introspection directly constitute the Specifically, Niedenthal et al. (2002) propose that
core conceptual content of social knowledge (Barsalou emotion congruence is determined by whether the tar-
et al., 2003). To understand the meaning of a trait or get stimulus must be simulated to produce the re-
stereotype is to have the ability to simulate the experi- sponse. During a mood induction, participants' affec-
ences of them competently. tive systems enter a particular emotional state (e.g.,
Consider the elderly stereotype and the associated happiness, sadness). When participants are subse-
trait of slow movement. According to embodied views, quently asked to judge a target that has particular emo-
amodal redescriptions of the modality-specific states tional properties (e.g., a happy person), they may rep-
associated with experiencing this stereotype and its as- resent the target with associated words (shallow
sociated trait do not represent them. Instead, knowl- processing), or they may construct a simulation of the
edge of elderly and slow movement resides in simula- target (deep processing). Because the simulation relies
tions of seeing and executing slow movements and, to on the same modality-specific systems that represent
some extent, in experiencing them in bodily states-no the induced emotion, a judge's current mood and the
further amodal descriptions of this trait are necessary. simulation have the potential to interact. For example,
Furthermore, theories such as PSS motivate novel when the judge is feeling happy and is judging a happy
predictions about when stereotypes will produce em- person, the mood and the target are consistent, such
bodiment effects (e.g., under deep, but not shallow, that the happiness judgment of the target is unusually
processing conditions) and what particular aspects of a high. Conversely, if the judge were sad when judging a
stereotype will be embodied on a particular occasion happy person, the mood and the target stimulus con-
(e.g., depending on what needs to be simulated in the flict, thereby lowering the judgment. Most important,
situated conceptualization). And embodiment theories if the target stimulus can be represented superficially
may also suggest factors that facilitate a revision of a with words-a simulation is unnecessary-the poten-
stereotype (e.g., change in the simulations strategy or tial for interaction between the mood and the stimulus
change in the modality-specific states underlying the representation does not exist, and mood congruency ef-
stereotype). fects do not occur. Such effects should only occur
when shared representational systems are used both to
represent the current mood and the target.
Attitude Change
In a similar fashion, theories of embodied cognition
can help us understand the learning and the unlearning Conclusion
of affectively charged cognition. In particular, such
theories may help us better understand social attitudes At first pass, theories of embodied cognition might
and prejudice, as well as the conditions that allow dif- seem a radical way to reconceptualize social mean-
ferent elements of an attitude to be expressed. Accord- ing. We believe, however, that these theories should
ing to PSS, the modality-specific states captured dur- be viewed not as threatening recent advances in so-
ing exposure to attitude objects constitute the core cial cognition, but as evolving from them toward com-
representation of the attitude. This implies, therefore, prehensive accounts of embodied phenomena that, tra-
that unlearning attitudes, including prejudicial ones, ditionally, have been difficult to explain. Similarly,
should involve more than a list of persuasive argu- other contemporary researchers argue that the central
ments. Interventions targeting the modality-specific problems of intersubjectivity (i.e., understanding oth-
bases of those concepts may also be useful. ers' intentions), empathy, and sympathy cannot be un-


derstood functionally if we do not take embodied ac- Barsalou, L. W. (2003a). Abstraction in perceptual symbol systems.
Philosophical Transactions ofthe Royal Society of London: Bi-
counts seriously (Gallese, 2003; Decety, in press). ological Sciences, 358, 1177-1187.
Furthermore, these notions are only radical if one ig- Barsalou, L. W. (2003b). Situated simulation in the human concep-
nores the many theoretical musings about the role of tual system. Language and Cognitive Processes, 18, 513-562.
embodiment in social interaction that preceded the Barsalou, L. W., Niedenthal, P. M., Barbey, A. K., & Ruppert, J. A.
dominance of the computer metaphor in cognitive psy- (2003). Social embodiment. In B. H. Ross (Ed.), The psychol-
chology (e.g., McDougall, 1908; Spencer, 1870). We ogy of learning and motivation, Vol. 43 (pp. 43-92). San Diego,
CA: Academic Press.
believe, therefore, that the theories of embodied cogni- Barsalou, L. W., Pecher, D., Zeelenberg, R., Simmons, W. K., &
tion can provide social psychologists with powerful Hamann, S. B. (in press). Multi-modal simulation in conceptual
new ways of theorizing about social representations processing. In W. Ahn, R. Goldstone, B. Love, A. Markman, &
and the mechanisms that process them. These theories P. Wolff (Eds.), Categorization inside and outside the lab:
also offer new ways of conceptualizing major social Festschrift in honor of Douglas L. Medin. Washington, DC:
American Psychological Association.
psychological constructs, such as attitudes, attitude Barsalou, L. W., Solomon, K. O., & Wu, L. L. (1999). Perceptual
change, impression formation, stereotyping, emotion, simulation in conceptual tasks. In M. K. Hiraga, C. Sinha, & S.
and empathy. Wilcox (Eds.), Cultural, typological, and psychological per-
spectives in cognitive linguistics: The proceedings of the 4th
conference of the International Cognitive Linguistics Associa-
tion, Vol. 3 (pp. 209-228). Amsterdam: John Benjamins.
References Barsalou, L. W., & Wiemer-Hastings, K. (in press). Situating ab-
stract concepts. In D. Pecher & R. Zwaan (Eds.), Grounding
Adolphs, R., Damasio, H., & Tranel, D. (2002). Neural systems for cognition: The role of perception and action in memory, lan-
recognition of emotional prosody: A 3-D lesion study. Emotion, guage, and thought. New York: Cambridge University Press.
2, 23-51. Bavelas, J. B., Black, A., Chovil, N., Lemery, C. R., & Mullett, J.
Adolphs, R., Damasio, H., Tranel, D., Cooper, G., & Damasio, A. R. (1988). Form and function in motor mimicry: Topographic evi-
(2000). A role for somatosensory cortices in the visual recogni- dence that the primary function is communicative. Human
tion of emotion as revealed by three dimensional lesion map- Communication Research, 14, 275-299.
ping. Journal ofNeuroscience, 20, 2683-2690. Bavelas, J. B., Black, A., Lemery, C. R., & Mullett, J. (1986). "I
Andersen, S. M., Reznik, I., & Manzella, L. M. (1996). Eliciting fa- show you how I feel": Motor mimicry as a communicative
cial affect, motivation, and expectancies in transference: Signif- act. Journal of Personality and Social Psychology, 50,
icant-other representations in social relations. Journal of Per- 322-329.
sonality and Social Psychology, 71, 1108-1129. Bernieri, F. (1988). Coordinated movement and rapport in teach-
Anderson, J. R. (1978). Arguments concerning representations for er-student interactions. Journal of Nonverbal Behavior, 12,
mental imagery. Psychological Review, 85, 249-277. 120-138.
Anderson, J. R. (1983). The architecture of cognition. Cambridge, Bernieri, F., Reznick, J. S., & Rosenthal, R. (1988). Synchrony,
MA: Harvard University Press. pseudo-synchrony, and dissynchrony: Measuring the entrain-
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: ment process in mother-infant interactions. Journal of Person-
Prentice-Hall. ality and Social Psychology, 54, 243-253.
Bargh, J. A., & Chartrand, T. L. (1999). The unbearable automaticity Bernieri, F., & Rosenthal, R. (1991). Interpersonal coordination: Be-
of being. American Psychologist, 54, 462-479. havioral matching and interactional synchrony. In R. S. Feld-
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social man & B. Rime (Eds.), Foundations of nonvervbal behavior
behavior: Direct effects of trait construct and stereotype activa- (pp. 401-432). New York: Cambridge University Press.
tion on action. Journal of Personality and Social Psychology, Berridge, K. C., & Winkielman, P. (2003). What is an unconscious
71, 230-244. emotion: The case for unconscious 'liking'. Cognition and
Barsalou, L. W. (1987). The instability of graded structure: Implica- Emotion, 17, 181-211.
tions for the nature of concepts. In U. Neisser (Ed.), Concepts Bless, H., & Fiedler, K. (1995). Affective states and the influence of
and conceptual development: Ecological and intellectual fac- activated general knowledge. Personality and Social Psychol-
tors in categorization (pp. 101-140). Cambridge, England: ogy Bulletin, 21, 766-778.
Cambridge University Press. Block, N. (1995). The mind as the software of the brain. In E. E.
Barsalou, L. W. (1989). Intraconcept similarity and its implications Smith & D. N. Osherson (Eds.), Thinking (pp. 377-425). Cam-
for interconcept similarity. In S. Vosniadou & A. Ortony (Eds.), bridge, MA: MIT Press.
Similarity and analogical reasoning (pp. 76-121). Cambridge, Bock, K. (1986). Syntactic persistence in sentence production. Cog-
England: Cambridge University Press. nitive Psychology, 18, 355-387.
Barsalou, L. W. (1993). Structure, flexibility, and linguistic vagary in Bower, G. H. (1981). Emotional mood and memory. American Psy-
concepts: Manifestations of a compositional system of percep- chologist, 36, 129- 148.
tual symbols. In A. C. Collins, S. E. Gathercole, & M. A. Breuer, J., & Freud, S. (1983-95/1955). Studies on hysteria. Lon-
Conway (Eds.), Theories of memory (pp. 29-101). London: don: Hogarth Press.
Lawrence Erlbaum Associates, Inc. Brooks, R. (1991). Intelligence without representation. Artificial In-
Barsalou, L. W. (1999). Perceptual symbol systems. Behavioral and telligence, 47, 139-159.
Brain Sciences, 22, 577-660. Bush, L., McHugo, G., & Lanzetta, J. (1989). The effects of sex and
Barsalou, L. W. (2002). Being there conceptually: Simulating cate- prior attitude on emotional reactions to expressive displays of
gories in preparation for situated action. In N. L. Stein, P. J. political leaders. Psychophysiology, 23, 427.
Bauer, & M. Rabinowitz (Eds.), Representation, memory, and Bush, L. K., Barr, C. L., McHugo, G. J., & Lanzetta, J. T. (1989). The
development: Essays in honor of Jean Mandler (pp. 1-15). effects of facial control and facial mimicry on subjective reac-
Mahwah, NJ: Lawrence Erlbaum Inc. tions to comedy routines. Motivation and Emotion, 13, 31-52.


Cacioppo, J. T., Priester, J. R., & Bernston, G. G. (1993). Rudimen- Decety, J. (in press). A social cognitive neuroscience model of hu-
tary determination of attitudes: II. Arm flexion and extension man empathy. In E. Harmon-Jones & P. Winkielman (Eds.),
have differential effects on attitudes. Journal of Personality and Fundamentals of Social Neuroscience. New York: Guilford.
Social Psychology, 65, 5-17. Decety, J., & Chaminade, T. (2003). Neural correlates of feeling
Cannon W. B. (1927). The James-Lange theory of emotions. Ameri- sympathy. Neuropsychologia. Special issue on Social Cogni-
can Journal of Psychology, 39, 115-124. tion, 41, 127-138.
Cannon, W. B. (1929). Bodily changes in pain, hunger; fear and Dennett, D. C. (1969). Content and consciousness. Oxford, England:
rage. New York: Appleton. Humanities Press.
Cappella, J. N., & Planalp, S. (1981). Talk and silence sequences in Deutsch, F. (1952). On the mysterious leap: From the mind to the
informal conversations III: Interspeaker influence. Human body. New York: International Universities Press.
Communication Research, 7, 117-132. Dijksterhuis, A., & Bargh, J. A. (2001). The perception-behavior ex-
Carlston, D. E. (1994). Associated systems theory: A systematic ap- pressway: Automatic effects of social perception on social be-
proach to the cognitive representation of persons and events. In havior. In M. P. Zanna (Ed.), Advances in experimental social
R. S. Wyer (Ed.), Advances in Social Cognition: Vol. 7. Associ- Psychology (Vol. 23, pp. 1-40). San Diego, CA: Academic Press.
ated Systems Theory (pp. 1-78). Hillsdale, NJ: Lawrence Erl- Dimberg, U. (1982). Facial reactions to facial expressions. Psy-
baum Associates, Inc. chophysiology, 19, 643-647.
Carr, L., lacoboni, M., Dubeau, M. C., Mazziotta, J. C., & Lenzi, G. Dimberg, U., Thunberg, M., & Elmehed, K. (2000). Unconscious fa-
L. (2003). Neural mechanisms of empathy in humans: A relay cial reactions to emotional facial expressions. Psychological
from neural systems for imitation to limbic areas. Proceedings Science, 11, 86-89.
of the National Academy of Science, 100, 5497-5502. Duclos, S. E., Laird, J. D., Schneider, E., Sexter, M., Stern, L., & Van
Chao, L. L., & Martin, A. (2000). Representation of manipulable Lighten, 0. (1989). Emotion-specific effects of facial expres-
man-made objects in the dorsal stream. Neuroimage, 12, sions and postures on emotional experience. Journal of Person-
478-484. ality and Social Psychology, 57, 100- 108.
Chartrand, T. I., & Bargh, J. A. (1999). The chameleon effect: The Eich, E., & Macaulay, D. (2000). Fundamental factors in mood de-
perception-behavior link and social interaction. Journal of Per- pendent memory. In J. P. Forgas (Ed.), Feeling and thinking:
sonality and Social Psychology, 76, 893-910. The role of affect in social cognition (pp. 109-130). New York:
Chen, S., & Bargh, J. A. (1999). Consequences of automatic evalua- Cambridge University Press.
tion: Immediate behavior predispositions to approach or avoid Farah, M. J. (1994). Neuropsychological inference with an interac-
the stimulus. Personality and Social Psychology Bulletin, 25, tive brain: A critique of the 'locality assumption.' Behavioral &
215-224. Brain Sciences, 17, 43-61.
Chin, P. S., & Ross, B. (2002). The effect of category learning on Farah, M. J. (2000). The neural bases of mental imagery. In M. S.
sensitivity to within-category correlations. Memory and Cogni- Gazzaniga (Ed.), The cognitive neurosciences (2nd ed., pp.
tion, 30, 353-362. 965-974). Cambridge, MA: MIT Press.
Chomsky, N. (1959). Review of B. F. Skinner's Verbal behavior. Farah, M. J., & McClelland, J. L. (1991). A computational model of
Language, 35, 26-58. semantic memory impairment: Modality specificity and emer-
Churchland, P. S., Ramachandran, V. S., & Sejnowski, T. J. (1994). A gent category specificity. Journal of Experimental Psychology:
critique of pure vision. In C. Koch & J. L. Davis (Eds.), Large- General, 120, 339-357.
scale neuronal theories of the brain. Computational neurosci- Finke, R. A. (1989). Principles ofmental imagery. Cambridge, MA:
ence (pp. 23-60). Cambridge, MA: MIT Press. MIT Press.
Clark, A. (1997). Being there. Putting brain, body, and world Fodor, J. A. (1975). The language of thought. New York: Crowell.
together again. Cambridge, MA: MIT Press. Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: MIT
Claypool, H. M., & Carlston, D. E. (2002). The effects of verbal and vi- Press.
sual interference on impressions: An associated systems approach. Forgas, J. P. (1995). Mood and judgment: The affect infusion model
Journal of Experimental Social Psychology, 38, 425-433. (AIM). Psychological Bulletin, 117, 39-66.
Collins, A. M., & Quillian, M. R. (1969). Retrieval time from seman- Forster, J., & Strack, F. (1997). Motor actions in retrieval of valenced
tic memory. Journal of Verbal learning and Verbal Behavior, 8, information: A motor congruence effect. Perceptual and Motor
240-247. Skills, 85, 1419-1427.
Cree, G. S., & McRae, K. (2003). Analyzing the factors underlying Forster, J., & Strack, F. (1998). Motor actions in retrieval of valenced
the structure and computation of the meaning of chipmunk, information: II. Boundary conditions for motor congruence ef-
cherry, chisel, cheese, and cello (and many other such concrete fects. Perceptual and Motor Skills, 86, 1423-1426.
nouns). Journal of Experimental Psychology: General, 132, Gallese, V. (2003). The manifold nature of interpersonal relations:
163-201. The quest for a common mechanism. Philosophical Transac-
Damasio, A. R. (1989). Time-locked multiregional retroactivation: tions of the Royal Society. B, 358: 517-528.
A systems-level proposal for the neural substrates of recall and Galton, F. (1884). Measurement of character. Fortnightly Review, 42,
recognition. Cognition, 33, 25-62. 179-185.
Damasio, A. R. (1994). Descartes'error: Emotion, reason, and the Glaser, W. R. (1992). Picture naming. Cognition, 42, 61-105.
human brain. New York: Grosset/Putnam. Glenberg, A. M. (1997). What memory is for. Behavioral and Brain
Damasio A. R. (1999). The feeling of what happens: Body and emo- Sciences, 20, 1-55.
tion in the making of consciousness. New York: Harcourt. Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in
Damasio A. R. (2003). Looking for Spinoza: Joy, sorrow, and the action. Psychonomic Bulletin & Review, 9, 558-565.
feeling brain. New York: Harcourt. Glenberg, A. M., & Robertson, D. A. (2000). Symbol grounding and
Damasio, A. R., & Damasio, H. (1994). Cortical systems for re- meaning: A comparison of high-dimensional and embodied
trieval of concrete knowledge: The convergence zone frame- theories of meaning. Journal of Memory and Language, 43,
work. In C. Koch & J. L. Davis (Eds.), Large-scale neuronal 379-401.
theories of the brain. Computational neuroscience (pp. 23-60). Graziano, W. G., Smith, S. M., Tassinary, L. G, Sun, C. R., &
Cambridge, MA: MIT Press. Pilkington, C. (1996). Does imitation enhance memory for
Darwin, C. (1872/1904). The expression of emotion in man and ani- faces? Four converging studies. Journal ofPersonality & Social
mals. London: Murray. Psychology, 71, 874-887.


Grezes, J., & Decety, J. (2001). Functional anatomy of execution, Markman, A. B., & Brendl, C. M. (in press). Constraining theories of
mental simulation, observation, and verb generation of actions: embodied cognition. Psychological Science.
A meta-analysis. Human Brain Mapping, 12, 1-19. Marques, J. F. (2004). Specialization and semantic organization: Ev-
Harnad, S. (2003). Symbol-grounding problem. Encyclopedia of idence for multiple-semantics as type ofknowledge. Manuscript
Cognitive Science. London: Nature Publishing Group. submitted for publication.
Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1992). Primitive emo- Martin, A. (2001). Functional neuroimaging of semantic memory.
tional contagion. In M. S. Clark (Ed.), Review of personality In. R. Cabeza & A. Kingstone (Eds.), Handbook ofjfunctional
and social psychology: Vol. 14: Emotions and social behavior neuroimaging of cognition (pp. 153-186). Cambridge, MA:
(pp. 151-177). Newbury Park, CA: Sage. MIT Press.
Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1993). Emotional con- Matarazzo, J. D., & Wiens, A. N. (1972). The interview: Research on
tagion. Current Directions in Psychological Science, 2, 96-99. its anatomy and structure. Chicago: Aldine-Atherton.
Heberlein, A. S., Adolphs, R., Tranel, D., & Damasio, H. (in press). Maxwell, G. M., Cook, M. W., & Burr, R. (1985). The encoding and
Cortical regions forjudgments ofemotions and personality traits decoding of liking from behavioral cues in both auditory and vi-
from point-light walkers. Journal of Cognitive Neuroscience. sual channels. Journal ofNonverbal Behavior, 9, 239-263.
Hegerty, M. (2004). Mechanical reasoning by mental simulation. McDougall, W. (1928). An introduction to social psychology.
Trends in Cognitive Sciences, 8, 280-285. Boston: John W. Luce. (Original work published in 1908)
Heidegger, M. (1962). Being and time. New York: Harper. McIntosh, D. N. (in press). Spontaneous facial mimicry, liking, and
Humphreys, G. W., & Forde, E. M. E. (2001). Hierarchies, similarity, emotional contagion. Polish Psychological Bulletin.
and interactivity in object recognition: "Category-specific" McIntosh, D. N., Druckman, D., &. Zajonc, R. B. (1994). So-
neuropsychological deficits. Behavioral & Brain Sciences, 24, cially-induced affect. In D. Druckman & R. A. Bjprk (Eds.),
453-509. Learning, remembering, believing: Enhancing human
Hutchison, W. D., Davis, K. D., Lozano, A. M., Tasker, R. R., & performance (pp. 251-276, 364-371). Washington, DC: Na-
Dostrovsky, J. 0. (1999). Pain-related neurons in the human tional Academy Press.
cingulate cortex. Nature-Neuroscience, 2, 403-405. McIntosh, D. N., Reichmann-Decker, A., Winkielman, P., & Wil-
lacoboni, M. (in press). Understanding others: Imitation, language, barger, J. (2004). When mirroring fails: Deficits in spontane-
empathy. In S. Hurley & N. Chater (Eds.), Perspectives on imi- ous, but not controlled mimicry ofemotionalfacial expressions
tation: From cognitive neuroscience to social science. Cam- in individuals with autism. Unpublished manuscript, University
bridge, MA: MIT Press. of Denver.
James, W. (1890). Principles of psychology (2 vols). New York: Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and man-
Holt, Rinehart & Winston. ual gestures by human neonates. Science, 198, 75-78.
James, W. (1994). The physical basis of emotion. Psychological Re- Meltzoff, A. N., & Moore, M. K. (1989). Imitation in newborn in-
view, 101, 205-210. (Original work published 1896) fants: Exploring the range of gestures imitated and the underly-
Kan, I. P., Barsalou, L. W., Solomon, K. O., Minor, J. K., & Thomp- ing mechanisms. Developmental Psychology, 25, 954-962.
son-Schill, S. L. (in press). Role of mental imagery in a property Meltzoff, A. N., & Prinz, W. (2002). The imitative mind. Develop-
verification task: fMRI evidence for perceptual representations ment, evolution, and brain bases. Cambridge studies in cogni-
of conceptual knowledge. Cognitive Neuropsychology. tive perceptual development. New York: Cambridge University
Kosslyn, S. M. (1980). Image and mind. Cambridge, MA: Harvard Press.
University Press. Merleau-Ponty, M. (1963). In praise of philosophy. Evanston, IL:
Kosslyn, S. M. (1994). Image and brain. Cambridge, MA: MIT Northwestern University Press.
Press. Neumann, R., Forster, J., & Strack, F. (2003). Motor compatibility:
Krauth-Gruber, S., Ric, F., & Niedenthal, P. M. (2004). The bidirectional link between behavior and evaluation. In J.
Kunda, Z. (1999). Social cognition: making sense ofpeople. Cam- Musch & K. C. Klauer (Eds.), The psychology of evaluation.
bridge, MA: MIT Press. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
LaFrance, M. (1985). Postural mirroring and intergroup relations. Neumann, R., & Strack, F. (2000). "Mood contagion": The auto-
Personality & Social Psychology Bulletin, 11, 207-217. matic transfer of mood between persons. Journal of Personality
LaFrance, M., & Ickes, W. (1981). Posture mirroring and inter- and Social Psychology, 79, 211-223.
actional involvement: Sex and sex typing effects. Journal of Newell, A. (1990). Unified theories of cognition. The William James
Nonverbal Behavior; 5, 139-154. lectures, 1987. Cambridge, MA: Harvard University Press.
Laird, J. D., Wagener, I. J., Halal, M., & Szegda, M. (1982). Remem- Newell, A., & Simon, H. A. (1972). Human problem solving. Ox-
bering what you feel: Effects of emotion on memory. Journal of ford, England: Prentice Hall.
Personality and Social Psychology, 42, 646-657. Newton, N. (1996). Foundations of understanding. Philadelphia:
Lakoff, G. (1997). Women, fire, and dangerous things: What catego- John Benjamins.
ries reveal about the mind. Chicago: University of Chicago Niedenthal, P. M. (1990). Implicit perception of affective in-
Press. formation. Journal of Experimental Social Psychology, 26,
Lakoff, G., & Johnson, M. (1 999). Philosophy in the flesh: The em- 505-527.
bodied mind and its challenge to western thought. New York: Niedenthal, P. M., Brauer, M., Halberstadt, J. B., & Innes-Ker, A. H.
Basic Books. (2001). When did her smile drop? Facial mimicry and the influ-
Lanzetta, J. T., & Englis, B. G. (1989). Expectations of cooperation ences of emotional state on the detection of change in emotional
and competition and their effects on observers' vicarious emo- expression. Cognition and Emotion, 15, 853-864.
tional responses. Journal ofPersonality and Social Psychology, Niedenthal, P. M., & Halberstadt, J. B. (2004). Top-down influences
56, 543-554. in social perception. In W. Stroebe & M. Hewstone (Eds.), Eu-
Levenson, R. W. (1996). Biological substrates of empathy and facial ropean Review of Social Psychology (Vol. 14, pp. 46-76). Lon-
modulation of emotion: Two facets of the scientific legacy of don: J. Wiley & Sons.
John Lanzetta. Motivation and Emotion, 20, 185-204. Niedenthal, P. M., Ric, F., & Krauth-Gruber, S. (2002). Explaining
l ipps, T. (1907). Das Wissen von fremden Ichen [The problem of emotion congruence and its absence in terms of perceptual sim-
the other's mind]. In T. Lipps (Ed.), Psychologische ulation. Psychological Inquiry, 13, 80-83.
Untersuchungen (Band 1, pp. 694-722). Leipzig, Germany: Niedenthal, P. M., Setterlund, M. B., & Jones, D. E. (1994). Emo-
Engelmann. tional organization of perceptual memory. In P. M. Niedenthal


& S. Kitayama (Eds.), The heart's eve: Emotional influences in Ryle, G. (1949). The concept of mind. Chicago: University of Chi-
perception and attention. New York: Academic Press. cago Press.
O'Reilly, R. C., & Munakata, Y. (2000). Computational explorations Schubert, T. W. (2004). The power in your hand: Gender differences
in cognitive neuroscience: Understanding the mind by simulat- in bodily feedback from making a fist. Personality and Social
ing the brain. Cambridge, MA: MIT Press. Psychology Bulletin, 30, 757-769.
O'Toole, R., & Dubin, R. (1968). Baby feeding and body sway: An Schwarz, N., Bless, H., Waenke, M., & Winkielman, P. (2003). Ac-
experiment in George Herbert Mead's "Taking the role of cessibility revisited. In G. V. Bodenhausen & A. J. Lambert
the other." Journal of Personality and Social Psychology, 10, (Eds.), Foundations ofsocial cognition: Afestschrift in honor of
59-65. Robert S. Wyer, Jr. (pp. 51-77). Hillsdale, NJ: Lawrence
Paivio, A. (1986). Mental representations: A dual coding approach. Erlbaum Associates, Inc.
New York: Oxford University Press. Schwarz, N., & Winkielman, P. (1999). Mood. In D. Levinson, J. J.
Palmer, S. E. (1999). Vision science: Photons to phenomenology. Ponzetti, Jr., & P. F. Jorgensen (Eds.), Encyclopedia of human
Cambridge, MA: MIT Press. emotions. (pp. 449-454). New York: Macmillan Reference.
Parsons, L., Fox, P., Downs, J., Glass, T., Hirsch, T., Martin, C., Searle, J. R. (1980). Mind, brains, and programs. The behavioral and
Jerabek, J., & Lancaster, J. (1995). Use of implicit motor imag- brain sciences, 3, 417-457.
ery for visual shape discrimination as revealed by PET. Nature, Semin, G. R. (2000). Agenda 2000- Communication: Language as
375, 54-58. an implementational device for cognition. European Journal of
Pecher, D., Zeelenberg, R., & Barsalou, L. (2003). Verifying concep- Social Psychology, 30, 595-612.
tual properties in different modalities produces switching costs. Shepard, R. N., & Cooper, L. A. (1982). Mental images and their
Psychological Science, 14, 119- 124. transformations. New York: Cambridge University Press.
Pecher, D., Zeelenberg, R., & Barsalou, L. W. (2004). Sensorimotor Sherrington, C. C. (1906). The integrative action of the nervous sys-
simulations underlie conceptual representations: Modality-spe- tem. New York: Yale University Press.
cific effects of prior activation. Psychonomic Bulletin & Re- Sigman, M. D., Kasari, C., Kwon, J.-h., & Yirmiya, N. (1992) Re-
view, 11, 164-167. sponses to the negative emotions of others by autistic, mentally
Piaget, J. (1972). The psychology of the child. New York: Basic retarded, and normal children. ChildDevelopment, 63, 796-807.
Books. Simmons, K., & Barsalou, L. W. (2003). The similarity-in-topogra-
Preston, S. D., & de Waal, F. B. M. (2002). Empathy: Its ultimate and phy principle: Reconciling theories of conceptual deficits. Cog-
proximate bases. Behavioral and Brain Sciences, 25, 1-71. nitive Neuropsychology, 20, 451486.
Prinz, J. J. (2002). Furnishing the mind: Concepts and their percep- Singer, T., Seymour, B., O'Doherty, J., Kaube, H., Dolan, R. J., &
tual basis. Cambridge, MA: MIT Press. Frith, C. D. (2004). Empathy for pain involves the affective but
Pylyshyn, Z. W. (1973). What the mind's eye tells the mind's brain: not sensory components of pain. Science, 303, 1157-1162.
A critique of mental imagery. Psychological Bulletin, 80, 1-24. Smith, E. R. (1998). Mental representation and memory. In D. T.
Pylyshyn, Z. W. (1981). The imagery debate: Analogue media ver- Gilbert, S. T. Fiske, & G. Lindzey-Gardner (Eds.). The hand-
sus tacit knowledge. Psychological Review, 88, 16-45. book of social psychology: Vol. 2 (4th ed., pp. 391-445). Bos-
Pylyshyn, Z. W. (1984). Computation and cognition: Toward a ton: McGraw-Hill.
foundation for cognitive science. Cambridge, MA: MIT Press. Smith, E. R., & Semin, G. R. (2004). Socially situated cognition:
Rauscher, F. H., Krauss, R. M., & Chen, Y. (1996). Gesture, speech, Cognition in its social context. Advances in Experimental So-
and lexical access: The role of lexical movements in speech cial Psychology, 36, 53-117.
production. Psychological Science, 7, 226-231. Solomon, K. O., & Barsalou, L. W. (2001). Representing properties
Reed, C. L., Farah, M. J. (1995). The psychological reality of the locally. Cognitive Psychology, 43, 129-169.
body schema: A test with normal participants. Journal of Ex- Solomon, K. O., & Barsalou, L. W. (2004). Perceptual simulation in
perimental Psychology: Human Perception & Performance, 21, property verification. Memory & Cognition, 32, 244-259.
334-343. Spencer, H. (1870). The principles of psychology. London: Williams
Richardson, D. C., Spivey, M. J., Barsalou, L. W., & McRae, K. and Norgate.
(2003). Spatial representations activated during real-time com- Spivey, M., Tyler, M., Richardson, D., & Young, E. (2000). Eye
prehension of verbs. Cognitive Science, 27, 767-780. movements during comprehension of spoken scene descrip-
Riskind, J. H. (1984). They stoop to conquer: Guiding and self-regu- tions. In Proceedings of the 22nd Annual Conference of the
latory functions of physical posture after success and failure. Cognitive Science Society, (pp. 487-492). Mahwah, NJ: Law-
Journal of Personality and Social Psychology, 47, 479-493. rence Erlbaum Associates, Inc.
Riskind, J. H., & Gotay, C. C. (1982). Physical posture: Could it have Stepper, S., & Strack, F. (1993). Proprioceptive determinants of
regulatory or feedback effects on motication and emotion? Mo- emotional and nonemotional feelings. Journal of Personality
tivation and Emotion, 6, 273- 298. and Social Psychology, 64, 211- 220.
Rizzolatti, G., & & Arbib, M. A. (1998). Language within our grasp. Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facili-
Trends in Neurosciences, 21, 188-194. tating conditions of the human smile: A nonobtrusive test of the
Rizzolatti, G., Fadiga, L., Fogassi, L., & Gallese, V. (2002). From facial feedback hypothesis. Journal of Personality and Social
mirror neurons to imitation: Facts and speculations. In A. N. Psychology, 54, 768-777.
Meltzoff & W. Prinz (Eds.), The imitative mind: Development, Tom, G., Pettersen, P., Lau, T., Burton, T., & Cook, J. (1991). The
evolution, and brain bases (pp. 247-266). New York: Cam- role of overt head movement in the formation of affect. Basic
bridge University Press. and Applied Social Psychology, 12, 281-289.
Rogers, S. (1999). An examination of the imitation deficit in autism. Tucker, M., & Ellis, R. (1998). On the relations between seen objects and
In J. Nadel & G. Butterworth (Eds.), Imitation in infancy (pp components of potential actions. Journal of Experimental Psychol-
255-283). Cambridge, MA: Cambridge University Press. ogy: Human Perception and Performance, 24, 830-846.
Rosch, E., & Mervis, C. B. (1975). Family resemblances: Studies in van Baaren, R. B., Maddux, W. W., Chartrand, T. L., de Bouter, C., &
the internal structure of categories. Cognitive Psychology, 7, van Knippenberg, A. (2003). It takes two to mimic: Behavioral
573-605. consequences of self-construals. Journal of Personality & So-
Rubin, D. C., & Greenberg, D. L. (1998). Visual memory deficit am- cial Psychology, 84, 1093- 1102.
nesia: A distinct amnesic presentation and etiology. Proceed- Vanman, E. J., Paul, B. Y., Ito, T. A., & Miller, N. (1997). The mod-
ings of the National Academy of Sciences, 95, 5413-5416. ern face of prejudice and structural features that moderate the


effect of cooperation on affect. Journal of Personality and So- Winkielman, P., Berntson, G. G., & Cacioppo, J. T. (2001). The
ciaC Psychology, 73, 941-959. psychophysiological perspective on the social mind. In A. Tes-
Varela, F. J., Thompson. E., & Rosch, E. (1991). The embodied ser & N. Schwarz (Eds.), Blackwell Handbook of Social Psy-
mind: Cognitive science and human experience. Cambridge, chology: Intraindividual Processes (pp. 89-108). Oxford, Eng-
MA: MIT Press. land: Blackwell.
Vaughan, K. B. & Lanzetta, J. T. (1980). Vicarious instigation and Winkielman, P., Berridge, K. C., & Wilbarger, J. (in press). Uncon-
conditioning of facial expressive and autonomic responses to a scious affective reactions to masked happy versus angry faces
model's expressive display of pain. Journal of Personality and influence consumption behavior and judgments of value. Per-
Social Psychologys 38, 909-923. sonality and Social Psychology Bulletin.
Wallbott, H. G. (1991). Recognition of emotion from facial expres- Winkielman, P., Zajonc, R. B., & Schwarz, N. (1997). Subliminal af-
sion via imitation? Some indirect evidence for an old theory. fective priming resists attributional interventions. Cognition
British Journal of Social Psychology, 30, 207-219. and Emotion, 11, 433-465.
Warrington, E. K., & Shallice, T. (1984). Category specific semantic Wu, L., & Barsalou, L. W. (2004). Perceptual simulation in property
impairments. Brain, 107, 829-854. generation. Manuscript submitted for publication.
Washburn, M. F. (1926). Movement and mental imagery: Outlines of Wyer, R. S., & Srull, T. K. (Eds.). (1984). Handbook ofsocial cogni-
a motor theory of the complexer mental processes. Boston: tion. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Houghlin Mifflin. Zajonc, R. B., Adelmann, P. K., Murphy, S. T., & Niedenthal, P. M.
Webb, J. T. (1972). Interview synchrony: An investigation of two (1987). Convergence in the physical appearance of spouses.
speech rate measures. In A. W. Seligman & B. Pope (Eds.), Motivation and Emotion, 11, 335-346.
Studies in dyadic communication (pp. 115-133). New York: Zajonc, R. B., & Markus, H. (1984). Affect and cognition: The hard
Pergamon. interface. In C. Izard, J. Kagan, & R. B. Zajonc (Eds.), Emo-
Wells, G. L., & Petty, R. E. (1980). The effects of overt head move- tions, cognition and behavior (pp. 73-102). Cambridge, Eng-
ments on persuasion: Compatibility and incompatibility of re- land: Cambridge University Press.
sponses. Basic and Applied Social Psychology, 1, 219-230. Zajonc, R. B., & McIntosh, D. N. (1992). Emotions research: Some
Wentura, D., Rothermund, K., & Bak, P. (2000). Automatic vigi- promising questions, some questionable promises. Psychologi-
lance: The attention-grabbing power of approach- and avoid- cal Science, 3, 70-74.
ance-related social information. Journal ofPersonality and So- Zajonc, R. B., Pietromonaco, P., & Bargh, J. (1982). Independence
cial Psychology, 78, 1024- 1037. and interaction of affect and cognition. In M. S. Clark & S.
Wicker, B., Keysers, C., Plailly, J., Royet, J.-P., Gallese, V., & Fiske (Eds.), Affect and Cognition: The 17th Annual Carnegie
Rizzolatti, G. (2003). Both of us disgusted in my insula: The Symposium on Cognition. Hillsdale, NJ: Lawrence Erlbaum
common neural basis of seeing and feeling disgust. Neuron, 40, Associates, Inc.
655-664. Zeki, S. (1993). A vision of the brain. Cambridge, MA: Blackwell.
Wilson, M. (2002). Six views of embodied cognition. Psychonomic
Bulletin & Review, 9, 625-636.